P
InmemoryofAmosTversky
P
PartII.HeuristicsandBiases
10.TheLawofSmallNumbers
<5>
11.Anchors
12.TheScienceofAvailability
13.Availability,Emotion,andRisk
14.TomW’sSpecialty
15.Linda:LessisMore
16.CausesTrumpStatistics
17.RegressiontotheMean
18.TamingIntuitivePredictions
PartIII.Overconfidence
19.TheIllusionofUnderstanding
20.TheIllusionofValidity
21.IntuitionsVs.Formulas
22.ExpertIntuition:WhenCanWeTrustIt?
23.TheOutsideView
24.TheEngineofCapitalism
PartIV.Choices
25.Bernoulli’sErrors
26.ProspectTheory
27.TheEndowmentEffect
28.BadEvents
29.TheFourfoldPattern
30.RareEvents
31.RiskPolicies
32.KeepingScore
33.Reversals
34.FramesandReality
PartV.TwoSelves
Introduction
Everyauthor,Isuppose,hasinmindasettinginwhichreadersofhisorherworkcould
benefitfromhavingreadit.Mineistheproverbialofficewatercooler,whereopinionsare
sharedandgossipisexchanged.Ihopetoenrichthevocabularythatpeopleusewhenthey
talk about the judgments and choices of others, the company’s new policies, or a
colleague’s investment decisions. Why be concerned with gossip? Because it is much
easier,aswellasfarmoreenjoyable,toidentifyandlabelthemistakesofothersthanto
recognizeourown.Questioningwhatwebelieveandwantisdifficultatthebestoftimes,
andespeciallydifficultwhenwemostneedtodoit,butwecanbenefitfromtheinformed
opinionsofothers.Manyofusspontaneouslyanticipatehowfriendsandcolleagueswill
evaluate our choices; the quality and content of these anticipated judgments therefore
matters. The expectation of intelligent gossip is a powerful motive for serious self-
criticism,morepowerfulthanNewYearresolutionstoimproveone’sdecisionmakingat
workandathome.
To be a good diagnostician, a physician needs to acquire a large set of labels for
diseases,eachofwhichbindsanideaoftheillnessanditssymptoms,possibleantecedents
andcauses,possibledevelopmentsandconsequences,andpossibleinterventionstocureor
mitigate the illness. Learning medicine consists in part of learning the language of
medicine. A deeper understanding of judgments and choices also requires a richer
vocabularythanisavailableineverydaylanguage.Thehopeforinformedgossipisthat
thereare distinctivepatterns inthe errorspeople make.Systematic errorsare knownas
biases, and they recur predictably in particular circumstances. When the handsome and
confidentspeakerboundsontothestage,forexample,youcananticipatethattheaudience
willjudgehiscommentsmorefavorablythanhedeserves.Theavailabilityofadiagnostic
label for this bias—the halo effect—makes it easier to anticipate, recognize, and
understand.
When you are asked what you are thinking about, you can normally answer. You
believe you know what goes on in your mind, which often consists of one conscious
thoughtleadinginanorderlywaytoanother.Butthatisnottheonlywaythemindworks,
norindeedisthatthetypicalway.Mostimpressionsandthoughtsariseinyourconscious
experience without your knowing how they got there. You cannot tracryd>e how you
cametothebeliefthatthereisalamponthedeskinfrontofyou,orhowyoudetecteda
hintofirritationinyourspouse’svoiceonthetelephone,orhowyoumanagedtoavoida
threat on the road before you became consciously aware of it. The mental work that
producesimpressions,intuitions,andmanydecisionsgoesoninsilenceinourmind.
Muchofthediscussioninthisbookisaboutbiasesofintuition.However,thefocus
onerrordoesnotdenigratehumanintelligence,anymorethantheattentiontodiseasesin
medicaltextsdeniesgoodhealth.Mostofusarehealthymostofthetime,andmostofour
judgments and actions are appropriate most of the time. As we navigate our lives, we
normallyallowourselvestobeguidedbyimpressionsandfeelings,andtheconfidencewe
haveinourintuitivebeliefsandpreferencesisusuallyjustified.Butnotalways.Weare
often confident even when we are wrong, and an objective observer is more likely to
detectourerrorsthanweare.
Sothisismyaimforwatercoolerconversations:improvetheabilitytoidentifyand
understand errors of judgment and choice, in others and eventually in ourselves, by
providingaricherandmorepreciselanguagetodiscussthem.Inatleastsomecases,an
accurate diagnosis may suggest an intervention to limit the damage that bad judgments
andchoicesoftencause.
Origins
Thisbookpresentsmycurrentunderstandingofjudgmentanddecisionmaking,whichhas
beenshapedbypsychologicaldiscoveriesofrecentdecades.However,Itracethecentral
ideastotheluckydayin1969whenIaskedacolleaguetospeakasaguesttoaseminarI
was teaching in the Department of Psychology at the Hebrew University of Jerusalem.
AmosTverskywasconsideredarisingstarinthefieldofdecisionresearch—indeed,in
anythinghedid—soIknewwewouldhaveaninterestingtime.Manypeoplewhoknew
Amos thought he was the most intelligent person they had ever met. He was brilliant,
voluble, and charismatic. He was also blessed with a perfect memory for jokes and an
exceptional ability to use them to make a point. There was never a dull moment when
Amoswasaround.Hewasthenthirty-two;Iwasthirty-five.
Amos told the class about an ongoing program of research at the University of
Michiganthatsoughttoanswerthisquestion:Arepeoplegoodintuitivestatisticians?We
alreadyknewthatpeoplearegoodintuitivegrammarians:atagefourachildeffortlessly
conformstotherulesofgrammarasshespeaks,althoughshehasnoideathatsuchrules
exist.Dopeoplehaveasimilarintuitivefeelforthebasicprinciplesofstatistics?Amos
reportedthattheanswerwasaqualifiedyes.Wehadalivelydebateintheseminarand
ultimatelyconcludedthataqualifiednowasabetteranswer.
Amos and I enjoyed the exchange and concluded that intuitive statistics was an
interestingtopicandthatitwouldbefuntoexploreittogether.ThatFridaywemetfor
lunchatCaféRimon,thefavoritehangoutofbohemiansandprofessorsinJerusalem,and
plannedastudyofthestatisticalintuitionsofsophisticatedresearchers.Wehadconcluded
in the seminar that our own intuitions were deficient. In spite of years of teaching and
using statistics, we had not developed an intuitive sense of the reliability of statistical
resultsobservedinsmallsamples.Oursubjectivejudgmentswerebiased:wewerefartoo
willingtobelieveresearchfindingsbasedoninadequateevidenceandpronetocollecttoo
fewobservationsinourownresearch.Thegoalofourstudywastoexaminewhetherother
researcherssufferedfromthesameaffliction.
Wepreparedasurveythatincludedrealisticscenariosofstatisticalissuesthatarisein
research.Amoscollectedtheresponsesofagroupofexpertparticipantsinameetingof
theSocietyofMathematicalPsychology,includingtheauthorsoftwostatisticaltextbooks.
As expected, we found that our expert colleagues, like us, greatly exaggerated the
likelihoodthattheoriginalresultofanexperimentwouldbesuccessfullyreplicatedeven
withasmallsample.Theyalsogaveverypooradvicetoafictitiousgraduatestudentabout
the number of observations she needed to collect. Even statisticians were not good
intuitivestatisticians.
Whilewritingthearticlethatreportedthesefindings,AmosandIdiscoveredthatwe
enjoyed working together. Amos was always very funny, and in his presence I became
funnyaswell,sowespenthoursofsolidworkincontinuousamusement.Thepleasurewe
foundin workingtogether made usexceptionally patient;itis mucheasier tostrive for
perfection when you are never bored. Perhaps most important, we checked our critical
weaponsatthedoor.BothAmosandIwerecriticalandargumentative,heevenmorethan
I,butduringtheyearsofourcollaborationneitherofuseverrejectedoutofhandanything
theothersaid.Indeed,oneofthegreatjoysIfoundinthecollaborationwasthatAmos
frequentlysawthepointofmyvagueideasmuchmoreclearlythanIdid.Amoswasthe
morelogicalthinker,withanorientationtotheoryandanunfailingsenseofdirection.I
wasmoreintuitiveandrootedinthepsychologyofperception,fromwhichweborrowed
manyideas.Weweresufficientlysimilartounderstandeachothereasily,andsufficiently
differenttosurpriseeachother.Wedevelopedaroutineinwhichwespentmuchofour
workingdaystogether,oftenonlongwalks.Forthenextfourteenyearsourcollaboration
wasthefocusofourlives,andtheworkwedidtogetherduringthoseyearswasthebest
eitherofuseverdid.
Wequicklyadoptedapracticethatwemaintainedformanyyears.Ourresearchwasa
conversation,inwhichweinventedquestionsandjointlyexaminedourintuitiveanswers.
Eachquestionwasasmallexperiment,andwecarriedoutmanyexperimentsinasingle
day.Wewerenotseriouslylookingforthecorrectanswertothestatisticalquestionswe
posed.Ouraimwastoidentifyandanalyzetheintuitiveanswer,thefirstonethatcameto
mind,theoneweweretemptedtomakeevenwhenweknewittobewrong.Webelieved
—correctly,asithappened—thatanyintuitionthatthetwoofussharedwouldbeshared
by many other people as well, and that it would be easy to demonstrate its effects on
judgments.
We once discovered with great delight that we had identical silly ideas about the
futureprofessionsofseveraltoddlerswebothknew.Wecouldidentifytheargumentative
three-year-old lawyer, the nerdy professor, the empathetic and mildly intrusive
psychotherapist. Of course these predictions were absurd, but we still found them
appealing.Itwasalsoclearthatourintuitionsweregovernedbytheresemblanceofeach
childtotheculturalstereotypeofaprofession.Theamusingexercisehelpedusdevelopa
theory that was emerging in our minds at the time, about the role of resemblance in
predictions.Wewentontotestandelaboratethattheoryindozensofexperiments,asin
thefollowingexample.
Asyouconsiderthenextquestion,pleaseassumethatStevewasselectedatrandom
fromarepresentativesample:
Anindividualhasbeendescribedbyaneighborasfollows:“Steveisveryshyand
withdrawn, invariably helpful but with little interest in people or in the world of
reality.Ameekandtidysoul,hehasaneedfororderandstructurutandstre,anda
passionfordetail.”IsStevemorelikelytobealibrarianorafarmer?
TheresemblanceofSteve’spersonalitytothatofastereotypicallibrarianstrikeseveryone
immediately,butequallyrelevantstatisticalconsiderationsarealmostalwaysignored.Did
it occur to you that there are more than 20 male farmers for each male librarian in the
United States? Because there are so many more farmers, it is almost certain that more
“meek and tidy” souls will be found on tractors than at library information desks.
However, we found that participants in our experiments ignored the relevant statistical
factsandreliedexclusivelyonresemblance.Weproposedthattheyusedresemblanceasa
simplifyingheuristic(roughly,aruleofthumb)tomakeadifficultjudgment.Thereliance
ontheheuristiccausedpredictablebiases(systematicerrors)intheirpredictions.
On another occasion, Amos and I wondered about the rate of divorce among
professorsinouruniversity.Wenoticedthatthequestiontriggeredasearchofmemoryfor
divorcedprofessorswekneworknewabout,andthatwejudgedthesizeofcategoriesby
the ease with which instances came to mind. We called this reliance on the ease of
memorysearchtheavailabilityheuristic.Inoneofourstudies,weaskedparticipantsto
answerasimplequestionaboutwordsinatypicalEnglishtext:
ConsidertheletterK.
IsKmorelikelytoappearasthefirstletterinawordORasthethirdletter?
AsanyScrabbleplayerknows,itismucheasiertocomeupwithwordsthatbeginwitha
particularletterthantofindwordsthathavethesameletterinthethirdposition.Thisis
trueforeveryletterofthealphabet.Wethereforeexpectedrespondentstoexaggeratethe
frequencyoflettersappearinginthefirstposition—eventhoseletters(suchasK,L,N,R,
V)whichinfactoccurmorefrequentlyinthethirdposition.Hereagain,therelianceona
heuristicproducesapredictablebiasinjudgments.Forexample,Irecentlycametodoubt
my long-held impression that adultery is more common among politicians than among
physiciansorlawyers.Ihadevencomeupwithexplanationsforthat“fact,”includingthe
aphrodisiac effect of power and the temptations of life away from home. I eventually
realizedthatthetransgressionsofpoliticiansaremuchmorelikelytobereportedthanthe
transgressions of lawyers and doctors. My intuitive impression could be due entirely to
journalists’choicesoftopicsandtomyrelianceontheavailabilityheuristic.
AmosandIspentseveralyearsstudyinganddocumentingbiasesofintuitivethinking
in various tasks—assigning probabilities to events, forecasting the future, assessing
hypotheses,andestimatingfrequencies.Inthefifthyearofourcollaboration,wepresented
ourmainfindingsinSciencemagazine,apublicationreadbyscholarsinmanydisciplines.
The article (which is reproduced in full at the end of this book) was titled “Judgment
Under Uncertainty: Heuristics and Biases.” It described the simplifying shortcuts of
intuitivethinkingandexplainedsome20biasesasmanifestationsoftheseheuristics—and
alsoasdemonstrationsoftheroleofheuristicsinjudgment.
Historiansofsciencehaveoftennotedthatatanygiventimescholarsinaparticular
fieldtendtosharebasicreshareassumptionsabouttheirsubject.Socialscientistsareno
exception; they rely on a view of human nature that provides the background of most
discussions of specific behaviors but is rarely questioned. Social scientists in the 1970s
broadlyacceptedtwoideasabouthumannature.First,peoplearegenerallyrational,and
their thinking is normally sound. Second, emotions such as fear, affection, and hatred
explain most of the occasions on which people depart from rationality. Our article
challengedbothassumptionswithoutdiscussingthemdirectly.Wedocumentedsystematic
errorsin thethinking of normalpeople, andwe tracedthese errors tothe designof the
machineryofcognitionratherthantothecorruptionofthoughtbyemotion.
Ourarticleattractedmuchmoreattentionthanwehadexpected,anditremainsoneof
themosthighlycitedworksinsocialscience(morethanthreehundredscholarlyarticles
referred to it in 2010). Scholars in other disciplines found it useful, and the ideas of
heuristics and biases have been used productively in many fields, including medical
diagnosis, legal judgment, intelligence analysis, philosophy, finance, statistics, and
militarystrategy.
Forexample,studentsofpolicyhavenotedthattheavailabilityheuristichelpsexplain
whysomeissuesarehighlysalientinthepublic’smindwhileothersareneglected.People
tendtoassesstherelativeimportanceofissuesbytheeasewithwhichtheyareretrieved
from memory—and this is largely determined by the extent of coverage in the media.
Frequentlymentionedtopicspopulatethemindevenasothersslipawayfromawareness.
Inturn,whatthemediachoosetoreportcorrespondstotheirviewofwhatiscurrentlyon
thepublic’smind.Itisnoaccidentthatauthoritarianregimesexertsubstantialpressureon
independentmedia.Becausepublicinterestismosteasilyarousedbydramaticeventsand
by celebrities, media feeding frenzies are common. For several weeks after Michael
Jackson’s death, for example, it was virtually impossible to find a television channel
reporting on another topic. In contrast, there is little coverage of critical but unexciting
issuesthatprovidelessdrama,suchasdecliningeducationalstandardsoroverinvestment
ofmedicalresourcesinthelastyearoflife.(AsIwritethis,Inoticethatmychoiceof
“little-covered”exampleswasguidedbyavailability.ThetopicsIchoseasexamplesare
mentioned often; equally important issues that are less available did not come to my
mind.)
We did not fully realize it at the time, but a key reason for the broad appeal of
“heuristics and biases” outside psychology was an incidental feature of our work: we
almostalwaysincludedinourarticlesthefulltextofthequestionswehadaskedourselves
and our respondents. These questions served as demonstrations for the reader, allowing
himtorecognizehowhisownthinkingwastrippedupbycognitivebiases.Ihopeyouhad
suchanexperienceasyoureadthequestionaboutStevethelibrarian,whichwasintended
tohelpyouappreciatethepowerofresemblanceasacuetoprobabilityandtoseehow
easyitistoignorerelevantstatisticalfacts.
The use of demonstrations provided scholars from diverse disciplines—notably
philosophersandeconomists—anunusualopportunitytoobservepossibleflawsintheir
own thinking. Having seen themselves fail, they became more likely to question the
dogmaticassumption,prevalentatthetime,thatthehumanmindisrationalandlogical.
The choice of method was crucial: if we had reported results of only conventional
experiments, the article would have been less noteworthy and less memorable.
Furthermore, skeptical readers would have distanced themselves from the results by
attributing judgment errors to the familiar l the famifecklessness of undergraduates, the
typicalparticipantsinpsychologicalstudies.Ofcourse,wedidnotchoosedemonstrations
overstandardexperimentsbecausewewantedtoinfluencephilosophersandeconomists.
We preferred demonstrations because they were more fun, and we were lucky in our
choiceofmethodaswellasinmanyotherways.Arecurrentthemeofthisbookisthat
luckplaysalargeroleineverystoryofsuccess;itisalmostalwayseasytoidentifyasmall
change in the story that would have turned a remarkable achievement into a mediocre
outcome.Ourstorywasnoexception.
Thereactiontoourworkwasnotuniformlypositive.Inparticular,ourfocusonbiases
wascriticizedassuggestinganunfairlynegativeviewofthemind.Asexpectedinnormal
science,someinvestigatorsrefinedourideasandothersofferedplausiblealternatives.By
and large, though, the idea that our minds are susceptible to systematic errors is now
generallyaccepted.Ourresearchonjudgmenthadfarmoreeffectonsocialsciencethan
wethoughtpossiblewhenwewereworkingonit.
Immediatelyaftercompletingourreviewofjudgment,weswitchedourattentionto
decision making under uncertainty. Our goal was to develop a psychological theory of
howpeoplemakedecisionsaboutsimplegambles.Forexample:Wouldyouacceptabet
onthetossofacoinwhereyouwin$130ifthecoinshowsheadsandlose$100ifitshows
tails? These elementary choices had long been used to examine broad questions about
decision making, such as the relative weight that people assign to sure things and to
uncertainoutcomes.Ourmethoddidnotchange:wespentmanydaysmakingupchoice
problems and examining whether our intuitive preferences conformed to the logic of
choice.Hereagain,asinjudgment,weobservedsystematicbiasesinourowndecisions,
intuitivepreferencesthatconsistentlyviolatedtherulesofrationalchoice.Fiveyearsafter
theSciencearticle,wepublished“ProspectTheory:AnAnalysisofDecisionUnderRisk,”
atheoryofchoicethatisbysomecountsmoreinfluentialthanourworkonjudgment,and
isoneofthefoundationsofbehavioraleconomics.
Untilgeographicalseparationmadeittoodifficulttogoon,AmosandIenjoyedthe
extraordinarygoodfortuneofasharedmindthatwassuperiortoourindividualmindsand
of a relationship that made our work fun as well as productive. Our collaboration on
judgmentanddecisionmakingwasthereasonfortheNobelPrizethatIreceivedin2002,
whichAmoswouldhavesharedhadhenotdied,agedfifty-nine,in1996.
Wherewearenow
ThisbookisnotintendedasanexpositionoftheearlyresearchthatAmosandIconducted
together,ataskthathasbeenablycarriedoutbymanyauthorsovertheyears.Mymain
aimhereistopresentaviewofhowthemindworksthatdrawsonrecentdevelopmentsin
cognitiveandsocialpsychology.Oneofthemoreimportantdevelopmentsisthatwenow
understandthemarvelsaswellastheflawsofintuitivethought.
Amos and I did not address accurate intuitions beyond the casual statement that
judgmentheuristics“arequiteuseful,butsometimesleadtosevereandsystematicerrors.”
We focused on biases, both because we found them interesting in their own right and
becausetheyprovidedevidencefortheheuristicsofjudgment.Wedidnotaskourselves
whether all intuitive judgments under uncertainty are produced by the heuristics we
studied;itisnowclearthattheyarenot.Inparticular,theaccurateintuitionsofexpertsare
betterexplainedbytheeffectsofprolongedpracticethanbyheuristics.Wecannowdraw
aricherandigharichemorebalancedpicture,inwhichskillandheuristicsarealternative
sourcesofintuitivejudgmentsandchoices.
The psychologist Gary Klein tells the story of a team of firefighters that entered a
houseinwhichthekitchenwasonfire.Soonaftertheystartedhosingdownthekitchen,
thecommanderheardhimselfshout,“Let’sgetoutofhere!”withoutrealizingwhy.The
floorcollapsedalmostimmediatelyafterthefirefightersescaped.Onlyafterthefactdid
thecommanderrealizethatthefirehadbeenunusuallyquietandthathisearshadbeen
unusually hot. Together, these impressions prompted what he called a “sixth sense of
danger.”Hehadnoideawhatwaswrong,butheknewsomethingwaswrong.Itturnedout
thattheheartofthefirehadnotbeeninthekitchenbutinthebasementbeneathwherethe
menhadstood.
Wehaveallheardsuchstoriesofexpertintuition:thechessmasterwhowalkspasta
streetgameandannounces“Whitematesinthree”withoutstopping,orthephysicianwho
makesacomplexdiagnosisafterasingleglanceatapatient.Expertintuitionstrikesusas
magical,butitisnot.Indeed,eachofusperformsfeatsofintuitiveexpertisemanytimes
eachday.Mostofusarepitch-perfectindetectingangerinthefirstwordofatelephone
call, recognize as we enter a room that we were the subject of the conversation, and
quicklyreacttosubtlesignsthatthedriverofthecarinthenextlaneisdangerous.Our
everyday intuitive abilities are no less marvelous than the striking insights of an
experiencedfirefighterorphysician—onlymorecommon.
The psychology of accurate intuition involves no magic. Perhaps the best short
statementofitisbythegreatHerbertSimon,whostudiedchessmastersandshowedthat
afterthousandsofhoursofpracticetheycometoseethepiecesontheboarddifferently
from the rest of us. You can feel Simon’s impatience with the mythologizing of expert
intuitionwhenhewrites:“Thesituationhasprovidedacue;thiscuehasgiventheexpert
accesstoinformationstoredinmemory,andtheinformationprovidestheanswer.Intuition
isnothingmoreandnothinglessthanrecognition.”
Wearenotsurprisedwhenatwo-year-oldlooksatadogandsays“doggie!”because
we are used to the miracle of children learning to recognize and name things.Simon’s
point is that the miracles of expert intuition have the same character. Valid intuitions
developwhenexpertshavelearnedtorecognizefamiliarelementsinanewsituationand
toactinamannerthatisappropriatetoit.Goodintuitivejudgmentscometomindwith
thesameimmediacyas“doggie!”
Unfortunately, professionals’ intuitions do not all arise from true expertise. Many
yearsagoIvisitedthechiefinvestmentofficerofalargefinancialfirm,whotoldmethat
hehadjustinvestedsometensofmillionsofdollarsinthestockofFordMotorCompany.
WhenIaskedhowhehadmadethatdecision,herepliedthathehadrecentlyattendedan
automobileshowandhadbeenimpressed.“Boy,dotheyknowhowtomakeacar!”was
hisexplanation.Hemadeitveryclearthathetrustedhisgutfeelingandwassatisfiedwith
himselfandwithhisdecision.Ifounditremarkablethathehadapparentlynotconsidered
the one question that an economist would call relevant: Is Ford stock currently
underpriced? Instead, he had listened to his intuition; he liked the cars, he liked the
company, and he liked the idea of owning its stock. From what we know about the
accuracyofstockpicking,itisreasonabletobelievethathedidnotknowwhathewas
doing.
The specific heuristics that Amos and I studied proviheitudied de little help in
understandinghowtheexecutivecametoinvestinFordstock,butabroaderconceptionof
heuristicsnowexists,whichoffersagoodaccount.Animportantadvanceisthatemotion
nowloomsmuchlargerinourunderstandingofintuitivejudgmentsandchoicesthanitdid
inthepast.Theexecutive’sdecisionwouldtodaybedescribedasanexampleoftheaffect
heuristic, where judgments and decisions are guided directly by feelings of liking and
disliking,withlittledeliberationorreasoning.
When confronted with a problem—choosing a chess move or deciding whether to
investinastock—themachineryofintuitivethoughtdoesthebestitcan.Iftheindividual
has relevant expertise, she will recognize the situation, and the intuitive solution that
comestohermindislikelytobecorrect.Thisiswhathappenswhenachessmasterlooks
atacomplexposition:thefewmovesthatimmediatelyoccurtohimareallstrong.When
thequestionisdifficultandaskilledsolutionisnotavailable,intuitionstillhasashot:an
answermaycometomindquickly—butitisnotananswertotheoriginalquestion.The
question that the executive faced (should I invest in Ford stock?) was difficult, but the
answertoaneasierandrelatedquestion(doIlikeFordcars?)camereadilytohismind
anddeterminedhischoice.Thisistheessenceofintuitiveheuristics:whenfacedwitha
difficult question, we often answer an easier one instead, usually without noticing the
substitution.
Thespontaneoussearchforanintuitivesolutionsometimesfails—neitheranexpert
solution nor a heuristic answer comes to mind. In such cases we often find ourselves
switching to a slower, more deliberate and effortful form of thinking. This is the slow
thinkingofthetitle.Fastthinkingincludesbothvariantsofintuitivethought—theexpert
and the heuristic—as well as the entirely automatic mental activities of perception and
memory,theoperationsthatenableyoutoknowthereisalamponyourdeskorretrieve
thenameofthecapitalofRussia.
The distinction between fast and slow thinking has been explored by many
psychologistsoverthelasttwenty-fiveyears.ForreasonsthatIexplainmorefullyinthe
nextchapter,Idescribementallifebythemetaphoroftwoagents,calledSystem1and
System2, whichrespectivelyproducefastandslowthinking.Ispeakofthe featuresof
intuitiveanddeliberatethoughtasiftheyweretraitsanddispositionsoftwocharactersin
yourmind.Inthepicturethatemergesfromrecentresearch,theintuitiveSystem1ismore
influentialthanyourexperiencetellsyou,anditisthesecretauthorofmanyofthechoices
andjudgmentsyoumake.MostofthisbookisabouttheworkingsofSystem1andthe
mutualinfluencesbetweenitandSystem2.
WhatComesNext
Thebookisdividedintofiveparts.Part1presentsthebasicelementsofatwo-systems
approach to judgment and choice. It elaborates the distinction between the automatic
operations of System 1 and the controlled operations of System 2, and shows how
associativememory,thecoreofSystem1,continuallyconstructsacoherentinterpretation
ofwhatisgoingoninourworldatanyinstant.Iattempttogiveasenseofthecomplexity
and richness of the automatic and often unconscious processes that underlie intuitive
thinking,andofhowtheseautomaticprocessesexplaintheheuristicsofjudgment.Agoal
istointroducealanguageforthinkingandtalkingaboutthemind.
Part2updatesthestudyofjudgmentheuristicsandexploresamajorpuzzle:Whyisit
sodifficultforustothinkstatistically?Weeasilythinkassociativelm1associay,wethink
metaphorically, we think causally, but statistics requires thinking about many things at
once,whichissomethingthatSystem1isnotdesignedtodo.
ThedifficultiesofstatisticalthinkingcontributetothemainthemeofPart3,which
describesapuzzlinglimitationofourmind:ourexcessiveconfidenceinwhatwebelieve
weknow,andourapparentinabilitytoacknowledgethefullextentofourignoranceand
the uncertainty of the world we live in. We are prone to overestimate how much we
understand about the world and to underestimate the role of chance in events.
Overconfidenceisfedbytheillusorycertaintyofhindsight.Myviewsonthistopichave
beeninfluencedbyNassimTaleb,theauthorofTheBlackSwan.Ihopeforwatercooler
conversationsthatintelligentlyexplorethelessonsthatcanbelearnedfromthepastwhile
resistingthelureofhindsightandtheillusionofcertainty.
Thefocusofpart4isaconversationwiththedisciplineofeconomicsonthenatureof
decisionmakingandontheassumptionthateconomicagentsarerational.Thissectionof
thebookprovidesacurrentview,informedbythetwo-systemmodel,ofthekeyconcepts
ofprospecttheory,themodelofchoicethatAmosandIpublishedin1979.Subsequent
chaptersaddressseveralwayshumanchoicesdeviatefromtherulesofrationality.Ideal
with the unfortunate tendency to treat problems in isolation, and with framing effects,
where decisions are shaped by inconsequential features of choice problems. These
observations, which are readily explained by the features of System 1, present a deep
challengetotherationalityassumptionfavoredinstandardeconomics.
Part5describesrecentresearchthathasintroducedadistinctionbetweentwoselves,
theexperiencingselfandtherememberingself,whichdonothavethesameinterests.For
example,wecanexposepeopletotwopainfulexperiences.Oneoftheseexperiencesis
strictlyworsethantheother,becauseitislonger.Buttheautomaticformationofmemories
—afeatureofSystem 1—has itsrules,whichwe can exploitsothattheworse episode
leaves a better memory. When people later choose which episode to repeat, they are,
naturally, guided by their remembering self and expose themselves (their experiencing
self) to unnecessary pain. The distinction between two selves is applied to the
measurementofwell-being,where we findagainthatwhat makestheexperiencingself
happyisnotquitethesameaswhatsatisfiestherememberingself.Howtwoselveswithin
asingle bodycan pursue happinessraises somedifficultquestions,both for individuals
andforsocietiesthatviewthewell-beingofthepopulationasapolicyobjective.
Aconcludingchapterexplores,inreverseorder,theimplicationsofthreedistinctions
drawn in the book: between the experiencing and the remembering selves, between the
conceptionofagentsinclassicaleconomicsandinbehavioraleconomics(whichborrows
from psychology), and between the automatic System 1 and the effortful System 2. I
returntothevirtuesofeducatinggossipandtowhatorganizationsmightdotoimprovethe
qualityofjudgmentsanddecisionsthataremadeontheirbehalf.
TwoarticlesIwrotewithAmosarereproducedasappendixestothebook.Thefirstis
thereviewofjudgmentunderuncertaintythatIdescribedearlier.Thesecond,publishedin
1984,summarizesprospecttheoryaswellasourstudiesofframingeffects.Thearticles
present the contributions that were cited by the Nobel committee—and you may be
surprisedbyhowsimpletheyare.Readingthemwillgiveyouasenseofhowmuchwe
knewalongtimeago,andalsoofhowmuchwehavelearnedinrecentdecades.
P
Part1
P
TwoSystems
P
TheCharactersoftheStory
Toobserveyourmindinautomaticmode,glanceattheimagebelow.
Figure1
Yourexperienceasyoulookatthewoman’sfaceseamlesslycombineswhatwenormally
call seeing and intuitive thinking. As surely and quickly as you saw that the young
woman’shairisdark,youknewsheisangry.Furthermore,whatyousawextendedinto
thefuture.Yousensedthatthiswomanisabouttosaysomeveryunkindwords,probably
inaloudandstridentvoice.Apremonitionofwhatshewasgoingtodonextcametomind
automaticallyandeffortlessly.Youdidnotintendtoassesshermoodortoanticipatewhat
shemightdo,andyourreactiontothepicturedidnothavethefeelofsomethingyoudid.
Itjusthappenedtoyou.Itwasaninstanceoffastthinking.
Nowlookatthefollowingproblem:
17×24
Youknewimmediatelythatthisisamultiplicationproblem,andprobablyknewthatyou
couldsolveit,withpaperandpencil,ifnotwithout.Youalsohadsomevagueintuitive
knowledge of the range of possible results. You would be quick to recognize that both
12,609and123areimplausible.Withoutspendingsometimeontheproblem,however,
youwouldnotbecertainthattheanswerisnot568.Aprecisesolutiondidnotcometo
mind,andyoufeltthatyoucouldchoosewhetherornottoengageinthecomputation.If
youhavenotdonesoyet,youshouldattemptthemultiplicationproblemnow,completing
atleastpartofit.
Youexperiencedslow thinkingas youproceeded through a sequence ofsteps. You
firstretrievedfrommemorythecognitiveprogramformultiplicationthatyoulearnedin
school,thenyouimplementedit.Carryingoutthecomputationwasastrain.Youfeltthe
burdenofholdingmuchmaterialinmemory,asyouneededtokeeptrackofwhereyou
were and of where you were going, while holding on to the intermediate result. The
processwasmentalwork:deliberate,effortful,andorderly—aprototypeofslowthinking.
Thecomputationwasnotonlyaneventinyourmind;yourbodywasalsoinvolved.Your
muscles tensed up, your blood pressure rose, and your heart rate increased. Someone
lookingcloselyatyoureyeswhileyoutackledthisproblemwouldhaveseenyourpupils
dilate.Yourpupilscontractedbacktonormalsizeassoonasyouendedyourwork—when
youfoundtheanswer(whichis408,bytheway)orwhenyougaveup.
TwoSystems
Psychologists have been intensely interested for several decades in the two modagee fi
Pn=“ceesofthinkingevokedbythepictureoftheangrywomanandbythemultiplication
problem,andhaveofferedmanylabelsforthem.Iadopttermsoriginallyproposedbythe
psychologists Keith Stanovich and Richard West, and will refer to two systems in the
mind,System1andSystem2.
System1operatesautomaticallyandquickly,withlittleornoeffortandnosenseof
voluntarycontrol.
System2allocatesattentiontotheeffortfulmentalactivitiesthatdemandit,including
complex computations. The operations of System 2 are often associated with the
subjectiveexperienceofagency,choice,andconcentration.
ThelabelsofSystem1andSystem2arewidelyusedinpsychology,butIgofurtherthan
mostinthisbook,whichyoucanreadasapsychodramawithtwocharacters.
Whenwethinkofourselves,weidentifywithSystem2,theconscious,reasoningself
thathasbeliefs,makeschoices,anddecideswhattothinkaboutandwhattodo.Although
System2believesitselftobewheretheactionis,theautomaticSystem1istheheroofthe
book.IdescribeSystem1aseffortlesslyoriginatingimpressionsandfeelingsthatarethe
main sources of the explicit beliefs and deliberate choices of System 2. The automatic
operations of System 1 generate surprisingly complex patterns of ideas, but only the
slower System 2 can construct thoughts in an orderly series of steps. I also describe
circumstances in which System 2 takes over, overruling the freewheeling impulses and
associationsofSystem1.Youwillbeinvitedtothinkofthetwosystemsasagentswith
theirindividualabilities,limitations,andfunctions.
Inroughorderofcomplexity,herearesomeexamplesoftheautomaticactivitiesthat
areattributedtoSystem1:
Detectthatoneobjectismoredistantthananother.
Orienttothesourceofasuddensound.
Completethephrase“breadand…”
Makea“disgustface”whenshownahorriblepicture.
Detecthostilityinavoice.
Answerto2+2=?
Readwordsonlargebillboards.
Driveacaronanemptyroad.
Findastrongmoveinchess(ifyouareachessmaster).
Understandsimplesentences.
Recognize that a “meek and tidy soul with a passion for detail” resembles an
occupationalstereotype.
All these mental events belong with the angry woman—they occur automatically and
requirelittleornoeffort.ThecapabilitiesofSystem1includeinnateskillsthatweshare
with other animals. We are born prepared to perceive the world around us, recognize
objects,orientattention,avoidlosses,andfearspiders.Othermentalactivitiesbecomefast
and automatic through prolonged practice. System 1 has learned associations between
ideas(thecapitalofFrance?);ithasalsolearnedskillssuchasreadingandunderstanding
nuancesofsocialsituations.Someskills,suchasfindingstrongchessmoves,areacquired
only by specialized experts. Others are widely shared. Detecting the similarity of a
personalitysketchtoanoccupatioheinoccupatnalstereotyperequiresbroadknowledgeof
the language and the culture, which most of us possess. The knowledge is stored in
memoryandaccessedwithoutintentionandwithouteffort.
Severalofthementalactionsinthelistarecompletelyinvoluntary.Youcannotrefrain
fromunderstanding simplesentences inyour ownlanguage orfrom orientingto aloud
unexpected sound, nor can you prevent yourself from knowing that 2 + 2 = 4 or from
thinking of Paris when the capital of France is mentioned. Other activities, such as
chewing, are susceptible to voluntary control but normally run on automatic pilot. The
controlofattentionissharedbythetwosystems.Orientingtoaloudsoundisnormallyan
involuntaryoperationofSystem1,whichimmediatelymobilizesthevoluntaryattention
ofSystem2.Youmaybeabletoresistturningtowardthesourceofaloudandoffensive
comment at a crowded party, but even if your head does not move, your attention is
initiallydirectedtoit,atleastforawhile.However,attentioncanbemovedawayfroman
unwantedfocus,primarilybyfocusingintentlyonanothertarget.
ThehighlydiverseoperationsofSystem2haveonefeatureincommon:theyrequire
attentionandaredisruptedwhenattentionisdrawnaway.Herearesomeexamples:
Braceforthestarterguninarace.
Focusattentionontheclownsinthecircus.
Focusonthevoiceofaparticularpersoninacrowdedandnoisyroom.
Lookforawomanwithwhitehair.
Searchmemorytoidentifyasurprisingsound.
Maintainafasterwalkingspeedthanisnaturalforyou.
Monitortheappropriatenessofyourbehaviorinasocialsituation.
Counttheoccurrencesoftheletterainapageoftext.
Tellsomeoneyourphonenumber.
Parkinanarrowspace(formostpeopleexceptgarageattendants).
Comparetwowashingmachinesforoverallvalue.
Filloutataxform.
Checkthevalidityofacomplexlogicalargument.
Inallthesesituationsyoumustpayattention,andyouwillperformlesswell,ornotatall,
ifyou arenot readyor ifyour attentionis directedinappropriately.System 2has some
ability to change the way System 1 works, by programming the normally automatic
functionsofattentionandmemory.Whenwaitingforarelativeatabusytrainstation,for
example,youcansetyourselfatwilltolookforawhite-hairedwomanorabeardedman,
andtherebyincreasethelikelihoodofdetectingyourrelativefromadistance.Youcanset
your memory to search for capital cities that start with N or for French existentialist
novels. And when you rent a car at London’s Heathrow Airport, the attendant will
probablyremindyou that“wedriveontheleftsideoftheroadoverhere.”Inallthese
cases,youareaskedtodosomethingthatdoesnotcomenaturally,andyouwillfindthat
theconsistentmaintenanceofasetrequirescontinuousexertionofatleastsomeeffort.
The often-used phrase “pay attention” is apt: you dispose of a limited budget of
attentionthatyoucanallocatetoactivities,andifyoutrytoi>Cyoutrytgobeyondyour
budget, you will fail. It is the mark of effortful activities that they interfere with each
other,whichiswhyitisdifficultorimpossibletoconductseveralatonce.Youcouldnot
compute the product of 17 × 24 while making a left turn into dense traffic, and you
certainlyshouldnottry.Youcandoseveralthingsatonce,butonlyiftheyareeasyand
undemanding.Youareprobablysafecarryingonaconversationwithapassengerwhile
drivingonanemptyhighway,andmanyparentshavediscovered,perhapswithsomeguilt,
thattheycanreadastorytoachildwhilethinkingofsomethingelse.
Everyone has some awareness of the limited capacity of attention, and our social
behaviormakesallowancesfortheselimitations.Whenthedriverofacarisovertakinga
truckonanarrowroad, forexample,adultpassengersquitesensiblystop talking. They
know that distracting the driver is not a good idea, and they also suspect that he is
temporarilydeafandwillnothearwhattheysay.
Intense focusing on a task can make people effectively blind, even to stimuli that
normallyattractattention.ThemostdramaticdemonstrationwasofferedbyChristopher
ChabrisandDanielSimonsintheirbookTheInvisibleGorilla.Theyconstructedashort
filmoftwoteamspassingbasketballs,oneteamwearingwhiteshirts,theotherwearing
black.Theviewersofthefilmareinstructedtocountthenumberofpassesmadebythe
white team, ignoring the black players. This task is difficult and completely absorbing.
Halfway through the video, a woman wearing a gorilla suit appears, crosses the court,
thumpsherchest,andmoveson.Thegorillaisinviewfor9seconds.Manythousandsof
peoplehaveseenthevideo,andabouthalfofthemdonotnoticeanythingunusual.Itis
thecountingtask—andespeciallytheinstructiontoignoreoneoftheteams—thatcauses
the blindness. No one who watches the video without that task would miss the gorilla.
Seeing and orienting are automatic functions of System 1, but they depend on the
allocation of some attention to the relevant stimulus. The authors note that the most
remarkableobservationoftheirstudyisthatpeoplefinditsresultsverysurprising.Indeed,
theviewerswhofailtoseethegorillaareinitiallysurethatitwasnotthere—theycannot
imagine missing such a striking event. The gorilla study illustrates two important facts
aboutourminds:wecanbeblindtotheobvious,andwearealsoblindtoourblindness.
PlotSynopsis
Theinteractionofthetwosystemsisarecurrentthemeofthebook,andabriefsynopsisof
theplotisinorder.InthestoryIwilltell,Systems1and2arebothactivewheneverwe
areawake.System1runsautomaticallyandSystem2isnormallyinacomfortablelow-
effortmode,inwhichonlyafractionof its capacity is engaged.System1continuously
generates suggestions for System 2: impressions, intuitions, intentions, and feelings. If
endorsedbySystem2,impressionsandintuitionsturnintobeliefs,andimpulsesturninto
voluntaryactions.Whenallgoessmoothly,whichismostofthetime,System2adoptsthe
suggestions of System 1 with little or no modification. You generally believe your
impressionsandactonyourdesires,andthatisfine—usually.
WhenSystem1runsintodifficulty,itcallsonSystem2tosupportmoredetailedand
specific processing that may solve the problem of the moment. System 2 is mobilized
whenaquestionarisesforwhichSystem1doesnotofferananswer,asprobablyhappened
to you when you encountered the multiplication problem 17 × 24. You can also feel a
surgeofconsciousattentionwheneveryouaresurprised.System2isactiv”><2isactated
whenaneventisdetectedthatviolatesthemodeloftheworldthatSystem1maintains.In
that world, lamps do not jump, cats do not bark, and gorillas do not cross basketball
courts. The gorilla experiment demonstrates that some attention is needed for the
surprisingstimulustobedetected.Surprisethenactivatesandorientsyourattention:you
willstare,andyouwillsearchyourmemoryforastorythatmakessenseofthesurprising
event.System2isalsocreditedwiththecontinuousmonitoringofyourownbehavior—
thecontrolthatkeepsyoupolitewhenyouareangry,andalertwhenyouaredrivingat
night.System2ismobilizedtoincreasedeffortwhenitdetectsanerrorabouttobemade.
Rememberatimewhenyoualmostblurtedoutanoffensiveremark andnotehowhard
youworkedtorestorecontrol.Insummary,mostofwhatyou(yourSystem2)thinkand
dooriginatesinyourSystem1,butSystem2takesoverwhenthingsgetdifficult,andit
normallyhasthelastword.
ThedivisionoflaborbetweenSystem1andSystem2ishighlyefficient:itminimizes
effortandoptimizesperformance.Thearrangementworkswellmostofthetimebecause
System 1 is generally very good at what it does: its models of familiar situations are
accurate,itsshort-termpredictionsareusuallyaccurateaswell,anditsinitialreactionsto
challengesareswiftandgenerallyappropriate.System1hasbiases,however,systematic
errorsthatitispronetomakeinspecifiedcircumstances.Asweshallsee,itsometimes
answerseasierquestionsthantheoneitwasasked,andithaslittleunderstandingoflogic
andstatistics.OnefurtherlimitationofSystem1isthatitcannotbeturnedoff.Ifyouare
shown a word on the screen in a language you know, you will read it—unless your
attentionistotallyfocusedelsewhere.
Conflict
Figure 2 is a variant of a classic experiment that produces a conflict between the two
systems.Youshouldtrytheexercisebeforereadingon.
Figure2
Youwerealmostcertainlysuccessfulinsayingthecorrectwordsinbothtasks,andyou
surelydiscoveredthatsomepartsofeachtaskweremucheasierthanothers.Whenyou
identifiedupper-andlowercase,theleft-handcolumnwaseasyandtheright-handcolumn
caused you to slow down and perhaps to stammer or stumble. When you named the
positionofwords,theleft-handcolumnwasdifficultandtheright-handcolumnwasmuch
easier.
ThesetasksengageSystem2,becausesaying“upper/lower”or“right/left”isnotwhat
youroutinelydowhenlookingdownacolumnofwords.Oneofthethingsyoudidtoset
yourselfforthetaskwastoprogramyourmemorysothattherelevantwords(upperand
lowerforthefirsttask)were“onthetipofyourtongue.”Theprioritizingofthechosen
wordsis effectiveandthe mildtemptationto read otherwords was fairlyeasy toresist
whenyouwentthroughthefirstcolumn.Butthesecondcolumnwasdifferent,becauseit
containedwordsforwhichyouwereset,andyoucouldnotignorethem.Youweremostly
able to respond correctly, but overcoming the competing response was a strain, and it
slowedyoudown.Youexperiencedaconflictbetweenataskthatyouintendedtocarry
outandanautomaticresponsethatinterferedwithit.
ConflictbetweenanautomaticreactionandanintentiontoconWhetiontoctrolitis
commoninourlives.Weareallfamiliarwiththeexperienceoftryingnottostareatthe
oddlydressedcoupleattheneighboringtableinarestaurant.Wealsoknowwhatitislike
toforceourattentiononaboringbook,whenweconstantlyfindourselvesreturningtothe
pointatwhichthereadinglostitsmeaning.Wherewintersarehard,manydrivershave
memoriesoftheircarskiddingoutofcontrolontheiceandofthestruggletofollowwell-
rehearsedinstructionsthatnegatewhattheywouldnaturallydo:“Steerintotheskid,and
whatever you do, do not touch the brakes!” And every human being has had the
experience of not telling someone to go to hell. One of the tasks of System 2 is to
overcometheimpulsesofSystem1.Inotherwords,System2isinchargeofself-control.
Illusions
ToappreciatetheautonomyofSystem1,aswellasthedistinctionbetweenimpressions
andbeliefs,takeagoodlookatfigure3.
This picture is unremarkable: two horizontal lines of different lengths, with fins
appended, pointing in different directions. The bottom line is obviously longer than the
oneaboveit.Thatiswhatweallsee,andwenaturallybelievewhatwesee.Ifyouhave
already encountered this image, however, you recognize it as the famous Müller-Lyer
illusion.Asyoucaneasilyconfirmbymeasuringthemwitharuler,thehorizontallines
areinfactidenticalinlength.
Figure3
Nowthatyouhavemeasuredthelines,you—yourSystem2,theconsciousbeingyou
call“I”—haveanewbelief:youknowthatthelinesareequallylong.Ifaskedabouttheir
length,youwillsaywhatyouknow.Butyoustillseethebottomlineaslonger.Youhave
chosentobelievethemeasurement,butyoucannotpreventSystem1fromdoingitsthing;
you cannot decide to see the lines as equal, although you know they are. To resist the
illusion,thereisonlyonethingyoucando:youmustlearntomistrustyourimpressionsof
thelengthoflineswhenfinsareattachedtothem.Toimplementthatrule,youmustbe
abletorecognizetheillusorypatternandrecallwhatyouknowaboutit.Ifyoucandothis,
youwillneveragainbefooledbytheMüller-Lyerillusion.Butyouwillstillseeoneline
aslongerthantheother.
Not all illusions are visual. There are illusions of thought, which we call cognitive
illusions. As a graduate student, I attended some courses on the art and science of
psychotherapy. During one of these lectures, our teacher imparted a morsel of clinical
wisdom.Thisiswhathetoldus:“Youwillfromtimetotimemeetapatientwhosharesa
disturbingtaleofmultiplemistakesinhisprevioustreatment.Hehasbeenseenbyseveral
clinicians, and all failed him. The patient can lucidly describe how his therapists
misunderstood him, but he has quickly perceived that you are different. You share the
samefeeling,areconvincedthatyouunderstandhim,andwillbeabletohelp.”Atthis
pointmyteacherraisedhisvoiceashesaid,“Donoteventhinkoftakingonthispatient!
Throwhimoutoftheoffice!Heismostlikelyapsychopathandyouwillnotbeableto
helphim.”
ManyyearslaterIlearnedthattheteacherhadwarnedusagainstpsychopathiccharm,
and the leading authority in the strn y in the udy of psychopathy confirmed that the
teacher’s advice was sound. The analogy to the Müller-Lyer illusion is close. What we
werebeingtaughtwasnothowtofeelaboutthatpatient.Ourteachertookitforgranted
thatthesympathywewouldfeelforthepatientwouldnotbeunderourcontrol;itwould
arisefromSystem1.Furthermore,wewerenotbeingtaughttobegenerallysuspiciousof
our feelings about patients. We were told that a strong attraction to a patient with a
repeatedhistoryoffailedtreatmentisadangersign—likethefinsontheparallellines.It
isanillusion—acognitiveillusion—andI(System2)wastaughthowtorecognizeitand
advisednottobelieveitoractonit.
Thequestionthatismostoftenaskedaboutcognitiveillusionsiswhethertheycanbe
overcome.Themessageoftheseexamplesisnotencouraging.BecauseSystem1operates
automaticallyandcannotbeturnedoffatwill,errorsofintuitivethoughtareoftendifficult
toprevent.Biasescannotalwaysbeavoided,becauseSystem2mayhavenocluetothe
error.Evenwhencuestolikelyerrorsareavailable,errorscanbepreventedonlybythe
enhanced monitoring and effortful activity of System 2. As a way to live your life,
however, continuous vigilance is not necessarily good, and it is certainly impractical.
Constantlyquestioningour ownthinkingwouldbeimpossiblytedious,andSystem2is
much too slow and inefficient to serve as a substitute for System 1 in making routine
decisions. The best we can do is a compromise: learn to recognize situations in which
mistakesarelikelyandtryhardertoavoidsignificantmistakeswhenthestakesarehigh.
Thepremiseofthisbookisthatitiseasiertorecognizeotherpeople’smistakesthanour
own.
UsefulFictions
Youhavebeeninvitedtothinkofthetwosystemsasagentswithinthemind,withtheir
individualpersonalities,abilities,andlimitations.Iwilloftenusesentencesinwhichthe
systemsarethesubjects,suchas,“System2calculatesproducts.”
The use of such language is considered a sin in the professional circles in which I
travel,becauseitseemstoexplainthethoughtsandactionsofapersonbythethoughts
and actions of little people inside the person’s head. Grammatically the sentence about
System2issimilarto“Thebutlerstealsthepettycash.”Mycolleagueswouldpointout
thatthebutler’saction actually explainsthedisappearance ofthecash, andtheyrightly
questionwhetherthesentenceaboutSystem2explainshowproductsarecalculated.My
answeristhatthebriefactivesentencethatattributescalculationtoSystem2isintended
asa description,not an explanation.It ismeaningful only becauseof whatyou already
knowaboutSystem2.Itisshorthandforthefollowing:“Mentalarithmeticisavoluntary
activity that requires effort, should not be performed while making a left turn, and is
associatedwithdilatedpupilsandanacceleratedheartrate.”
Similarly, the statement that “highway driving under routine conditions is left to
System1”meansthatsteeringthecararoundabendisautomaticandalmosteffortless.It
alsoimpliesthatanexperienceddrivercandriveonanemptyhighwaywhileconductinga
conversation. Finally, “System 2 prevented James from reacting foolishly to the insult”
means that James would have been more aggressive in his response if his capacity for
effortfulcontrolhadbeendisrupted(forexample,ifhehadbeendrunk).
System1andSystem2aresocentraltothestoryItellinthisbookthatImustmakeit
absolutelyclear thatthey are217at they a fictitiouscharacters. Systems1 and 2are not
systemsinthestandardsenseofentitieswithinteractingaspectsorparts.Andthereisno
onepartofthebrainthateitherofthesystemswouldcallhome.Youmaywellask:What
isthepointofintroducingfictitiouscharacterswithuglynamesintoaseriousbook?The
answeristhatthecharactersareusefulbecauseofsomequirksofourminds,yoursand
mine.Asentenceisunderstoodmoreeasilyifitdescribeswhatanagent(System2)does
thanifitdescribeswhatsomethingis,whatpropertiesithas.Inotherwords,“System2”is
abettersubjectforasentencethan“mentalarithmetic.”Themind—especiallySystem1—
appearstohaveaspecialaptitudefortheconstructionandinterpretationofstoriesabout
active agents, who have personalities, habits, and abilities. You quickly formed a bad
opinion of the thieving butler, you expect more bad behavior from him, and you will
rememberhimforawhile.Thisisalsomyhopeforthelanguageofsystems.
Why call them System 1 and System 2 rather than the more descriptive “automatic
system”and“effortfulsystem”?Thereasonissimple:“Automaticsystem”takeslongerto
say than “System 1” and therefore takes more space in your working memory. This
matters, because anything that occupies your working memory reduces your ability to
think. You should treat “System 1” and “System 2” as nicknames, like Bob and Joe,
identifyingcharactersthatyouwillgettoknowoverthecourseofthisbook.Thefictitious
systemsmakeiteasierformetothinkaboutjudgmentandchoice,andwillmakeiteasier
foryoutounderstandwhatIsay.
SpeakingofSystem1andSystem2
“Hehadanimpression,butsomeofhisimpressionsareillusions.”
“ThiswasapureSystem1response.Shereactedtothethreatbeforesherecognized
it.”
“ThisisyourSystem1talking.SlowdownandletyourSystem2takecontrol.”
P
AttentionandEffort
Intheunlikelyeventofthisbookbeingmadeintoafilm,System2wouldbeasupporting
characterwhobelievesherselfto bethehero.Thedefining feature ofSystem2,in this
story,isthatitsoperationsareeffortful,andoneofitsmaincharacteristicsislaziness,a
reluctancetoinvestmoreeffortthanisstrictlynecessary.Asaconsequence,thethoughts
and actions that System 2 believes it has chosen are often guided by the figure at the
center of the story, System 1. However, there are vital tasks that only System 2 can
perform because they require effort and acts of self-control in which the intuitions and
impulsesofSystem1areovercome.
MentalEffort
IfyouwishtoexperienceyourSystem2workingatfulltilt,thefollowingexercisewill
do; it should br”0%e ca Tting you to the limits of your cognitive abilities within 5
seconds.Tostart,makeupseveralstringsof4digits,alldifferent,andwriteeachstringon
anindexcard.Placeablankcardontopofthedeck.Thetaskthatyouwillperformis
calledAdd-1.Hereishowitgoes:
Startbeatingasteadyrhythm(orbetteryet,setametronomeat1/sec).Removethe
blankcardandreadthefourdigitsaloud.Waitfortwobeats,thenreportastringin
whicheach of the original digitsis incrementedby 1. If the digitson thecard are
5294,thecorrectresponseis6305.Keepingtherhythmisimportant.
FewpeoplecancopewithmorethanfourdigitsintheAdd-1task,butifyouwantaharder
challenge,pleasetryAdd-3.
Ifyouwouldliketoknowwhatyourbodyisdoingwhileyourmindishardatwork,
setuptwopilesofbooksonasturdytable,placeavideocameraononeandleanyour
chinon theother, getthe videogoing, andstare atthe cameralens whileyou workon
Add-1 or Add-3 exercises. Later, you will find in the changing size of your pupils a
faithfulrecordofhowhardyouworked.
IhavealongpersonalhistorywiththeAdd-1task.EarlyinmycareerIspentayearat
the University of Michigan, as a visitor in a laboratory that studied hypnosis. Casting
aboutforausefultopicofresearch,IfoundanarticleinScientificAmericaninwhichthe
psychologistEckhardHessdescribedthepupiloftheeyeasawindowtothesoul.Ireread
it recently and again found it inspiring. It begins with Hess reportingthat his wife had
noticedhispupilswideningashewatchedbeautifulnaturepictures,anditendswithtwo
striking pictures of the same good-looking woman, who somehow appears much more
attractive in one than in the other. There is only one difference: the pupils of the eyes
appear dilated in the attractive picture and constricted in the other. Hess also wrote of
belladonna,apupil-dilatingsubstancethatwasusedasacosmetic,andofbazaarshoppers
whoweardarkglassesinordertohidetheirlevelofinterestfrommerchants.
One of Hess’s findings especially captured my attention. He had noticed that the
pupils are sensitive indicators of mental effort—they dilate substantially when people
multiplytwo-digitnumbers,andtheydilatemoreiftheproblemsarehardthaniftheyare
easy. His observations indicated that the response to mental effort is distinct from
emotionalarousal.Hess’sworkdidnothavemuchtodowithhypnosis,butIconcluded
thatthe idea of a visibleindication ofmentalefforthadpromise asa researchtopic. A
graduatestudentinthelab,JacksonBeatty,sharedmyenthusiasmandwegottowork.
BeattyandIdevelopedasetupsimilartoanoptician’sexaminationroom,inwhich
theexperimentalparticipantleanedherheadonachin-and-foreheadrestandstaredata
camera while listening to prerecorded information and answering questions on the
recorded beats of a metronome. The beats triggered an infrared flash every second,
causingapicturetobetaken.Attheendofeachexperimentalsession,wewouldrushto
havethefilmdeveloped,projecttheimagesofthepupilonascreen,andgotoworkwitha
ruler. The method was a perfect fit for young and impatient researchers: we knew our
resultsalmostimmediately,andtheyalwaystoldaclearstory.
BeattyandIfocusedonpacedtasks,suchasAdd-1,inwhichweknewpreciselywhat
was on the subject’s mind at any time. We recorded strings of digits on beats of the
metronomeandinstructedthesubjecttorepeatortransformthedigitsoneindigitsonby
one,maintainingthesamerhythm.Wesoondiscoveredthatthesizeofthepupilvaried
secondbysecond,reflectingthechangingdemandsofthetask.Theshapeoftheresponse
wasaninverted V.Asyou experienceditifyou triedAdd-1or Add-3, effortbuilds up
witheveryadded digitthatyou hear, reachesan almostintolerablepeak as yourush to
produce a transformed string during and immediately after the pause, and relaxes
graduallyasyou“unload”yourshort-termmemory.Thepupildatacorrespondedprecisely
tosubjectiveexperience:longerstringsreliablycausedlargerdilations,thetransformation
taskcompoundedtheeffort,andthepeakofpupilsizecoincidedwithmaximumeffort.
Add-1withfourdigitscausedalargerdilationthanthetaskofholdingsevendigitsfor
immediaterecall.Add-3,whichismuchmoredifficult,isthemostdemandingthatIever
observed.Inthe first5seconds, the pupildilatesby about 50%of its originalareaand
heartrateincreasesbyabout7beatsperminute.Thisisashardaspeoplecanwork—they
giveupifmoreisaskedofthem.Whenweexposedoursubjectstomoredigitsthanthey
couldremember,theirpupilsstoppeddilatingoractuallyshrank.
Weworkedforsomemonthsinaspaciousbasementsuiteinwhichwehadsetupa
closed-circuit system that projected an image of the subject’s pupil on a screen in the
corridor;wealsocouldhearwhatwashappeninginthelaboratory.Thediameterofthe
projectedpupilwasaboutafoot;watchingitdilateandcontractwhentheparticipantwas
at work was a fascinating sight, quite an attraction for visitors in our lab. We amused
ourselvesandimpressedourguestsbyourabilitytodivinewhentheparticipantgaveup
onatask.Duringamentalmultiplication,thepupilnormallydilatedtoalargesizewithin
afewsecondsandstayedlargeaslongastheindividualkeptworkingontheproblem;it
contractedimmediatelywhenshefoundasolutionorgaveup.Aswewatchedfromthe
corridor, we would sometimes surprise both the owner of the pupil and our guests by
asking,“Whydidyoustopworkingjustnow?”Theanswerfrominsidethelabwasoften,
“Howdidyouknow?”towhichwewouldreply,“Wehaveawindowtoyoursoul.”
Thecasualobservationswemadefromthecorridorweresometimesasinformativeas
theformalexperiments.ImadeasignificantdiscoveryasIwasidlywatchingawoman’s
pupilduringabreakbetweentwotasks.Shehadkeptherpositiononthechinrest,soI
could see the image of her eye while she engaged in routine conversation with the
experimenter.Iwassurprisedtoseethatthepupilremainedsmallanddidnotnoticeably
dilate as she talked and listened. Unlike the tasks that we were studying, the mundane
conversationapparentlydemandedlittleornoeffort—nomorethanretainingtwoorthree
digits.Thiswasaeurekamoment:Irealizedthatthetaskswehadchosenforstudywere
exceptionallyeffortful.Animagecametomind:mentallife—todayIwouldspeakofthe
life of System 2—is normally conducted at the pace of a comfortable walk, sometimes
interruptedbyepisodesofjoggingandonrareoccasionsbyafranticsprint.TheAdd-1
andAdd-3exercisesaresprints,andcasualchattingisastroll.
We found that people, when engaged in a mental sprint, may become effectively
blind.TheauthorsofTheInvisibleGorillahadmadethegorilla“invisible”bykeepingthe
observersintenselybusycountingpasses.Wereportedaratherlessdramaticexampleof
blindnessduringAdd-1.Oursubjectswereexposedtoaseriesofrapidlyflashingletters
whiletheyworked.Theyweretoldtogivethetaskcompletepriority,buttheywerealso
askedtoreport,attheendofthedigittask,whethertheletterKhadappearedatanyrored
atantimeduringthetrial.Themainfindingwasthattheabilitytodetectandreportthe
targetletterchangedinthecourseofthe10secondsoftheexercise.Theobserversalmost
nevermissedaKthatwasshownatthebeginningorneartheendoftheAdd-1taskbut
theymissedthetargetalmosthalfthetimewhenmentaleffortwasatitspeak,althoughwe
hadpicturesoftheirwide-openeyestaringstraightatit.Failuresofdetectionfollowedthe
sameinverted-Vpatternasthedilatingpupil.Thesimilaritywasreassuring:thepupilwas
agoodmeasureofthephysicalarousalthataccompaniesmentaleffort,andwecouldgo
aheadanduseittounderstandhowthemindworks.
Muchliketheelectricitymeteroutsideyourhouseorapartment,thepupilsofferan
indexofthecurrentrateatwhichmentalenergyisused.Theanalogygoesdeep.Youruse
ofelectricitydependsonwhatyouchoosetodo,whethertolightaroomortoastapieceof
bread.Whenyou turnona bulbora toaster, itdrawsthe energyitneeds but nomore.
Similarly,wedecidewhattodo,butwehavelimitedcontrolovertheeffortofdoingit.
Supposeyouareshownfourdigits,say,9462,andtoldthatyourlifedependsonholding
them in memory for 10 seconds. However much you want to live, you cannot exert as
much effort in this task as you would be forced to invest to complete an Add-3
transformationonthesamedigits.
System2andtheelectricalcircuitsinyourhomebothhavelimitedcapacity,butthey
responddifferentlytothreatenedoverload.Abreakertripswhenthedemandforcurrentis
excessive, causing all devices on that circuit to lose power at once. In contrast, the
responsetomentaloverloadisselectiveandprecise:System2protectsthemostimportant
activity, so it receives the attention it needs; “spare capacity” is allocated second by
second to other tasks. In our version of the gorilla experiment, we instructed the
participants to assign priority to the digit task. We know that they followed that
instruction,becausethetimingofthevisualtargethadnoeffectonthemaintask.Ifthe
criticalletterwaspresentedatatimeofhighdemand,thesubjectssimplydidnotseeit.
Whenthetransformationtaskwaslessdemanding,detectionperformancewasbetter.
The sophisticated allocation of attention has been honed by a long evolutionary
history. Orienting and responding quickly to the gravest threats or most promising
opportunities improved the chance of survival, and this capability is certainly not
restrictedto humans. Even inmodernhumans,System1takesoverinemergenciesand
assignstotalprioritytoself-protectiveactions.Imagineyourselfatthewheelofacarthat
unexpectedlyskidsonalargeoilslick.Youwillfindthatyouhaverespondedtothethreat
beforeyoubecamefullyconsciousofit.
BeattyandIworkedtogetherforonlyayear,butourcollaborationhadalargeeffect
on our subsequent careers. He eventually became the leading authority on “cognitive
pupillometry,”and Iwrotea book titledAttention and Effort,whichwasbased in large
partonwhatwelearnedtogetherandonfollow-upresearchIdidatHarvardthefollowing
year.Welearnedagreatdealabouttheworkingmind—whichInowthinkofasSystem2
—frommeasuringpupilsinawidevarietyoftasks.
As you become skilled in a task, its demand for energy diminishes. Studies of the
brain have shown that the pattern of activity associated with an action changes as skill
increases,withfewerbrainregionsinvolved.Talenthassimilareffects.Highlyintelligent
individualsneedlessefforttosolvethesameproblems,asindicatedbybothpupilsizeand
brainactivity.Ageneral“lawofleasteffort”appdt”aliestocognitiveaswellasphysical
exertion.Thelawassertsthatifthereareseveralwaysofachievingthesamegoal,people
will eventually gravitate to the least demanding course of action. In the economy of
action,effortisacost,andtheacquisitionofskillisdrivenbythebalanceofbenefitsand
costs.Lazinessisbuiltdeepintoournature.
Thetasksthatwestudiedvariedconsiderablyintheireffectsonthepupil.Atbaseline,
oursubjectswereawake,aware,andreadytoengageinatask—probablyatahigherlevel
of arousal and cognitive readiness than usual. Holding one or two digits in memory or
learning to associate a word with a digit (3 = door) produced reliable effects on
momentaryarousalabove thatbaseline,but theeffectswere minuscule,only5% ofthe
increase in pupil diameter associated with Add-3. A task that required discriminating
betweenthepitchoftwotonesyieldedsignificantlylargerdilations.Recentresearchhas
shown that inhibiting the tendency to read distracting words (as in figure 2 of the
preceding chapter) also induces moderate effort. Tests of short-term memory for six or
sevendigitsweremoreeffortful.Asyoucanexperience,therequesttoretrieveandsay
aloudyourphonenumberoryourspouse’sbirthdayalso requires a briefbutsignificant
effort,becausetheentirestringmustbeheldinmemoryasaresponseisorganized.Mental
multiplicationoftwo-digitnumbers and theAdd-3task are nearthelimit of whatmost
peoplecando.
What makes some cognitive operations more demanding and effortful than others?
Whatoutcomesmustwepurchaseinthecurrencyofattention?WhatcanSystem2dothat
System1cannot?Wenowhavetentativeanswerstothesequestions.
Effort is required to maintain simultaneously in memory several ideas that require
separate actions, or that need to be combined according to a rule—rehearsing your
shoppinglistasyou enter thesupermarket,choosingbetweenthe fish andthevealat a
restaurant,or combininga surprisingresult froma surveywith theinformation thatthe
samplewassmall,forexample.System2istheonlyonethatcanfollowrules,compare
objectsonseveralattributes,andmakedeliberatechoicesbetweenoptions.Theautomatic
System1doesnothavethesecapabilities.System1detectssimplerelations(“theyareall
alike,”“thesonismuchtallerthanthefather”)andexcelsatintegratinginformationabout
onething,butitdoesnotdealwithmultipledistincttopicsatonce,norisitadeptatusing
purelystatisticalinformation.System1willdetectthatapersondescribedas“ameekand
tidy soul, with a need for order and structure, and a passion for detail” resembles a
caricaturelibrarian,butcombiningthisintuitionwithknowledgeaboutthesmallnumber
oflibrariansisataskthatonlySystem2canperform—ifSystem2knowshowtodoso,
whichistrueoffewpeople.
AcrucialcapabilityofSystem2istheadoptionof“tasksets”:itcanprogrammemory
toobeyaninstructionthatoverrideshabitualresponses.Considerthefollowing:Countall
occurrencesoftheletterfinthispage.Thisisnotataskyouhaveeverperformedbefore
anditwillnotcomenaturallytoyou,butyourSystem2cantakeiton.Itwillbeeffortful
to set yourself up for this exercise, and effortful to carry it out, though you will surely
improvewithpractice.Psychologistsspeakof“executivecontrol”todescribetheadoption
andterminationof task sets, and neuroscientistshaveidentifiedthemainregionsofthe
brain that serve the executive function. One of these regions is involved whenever a
conflict must be resolved. Another is the prefrontal area of the brain, a region that is
substantially more developed in humans tht un humans an in other primates, and is
involvedinoperationsthatweassociatewithintelligence.
Now suppose that at the end of the page you get another instruction: count all the
commas in the next page. This will be harder, because you will have to overcome the
newly acquired tendency to focus attention on the letter f. One of the significant
discoveriesofcognitivepsychologistsinrecentdecadesisthatswitchingfromonetaskto
anotheriseffortful,especiallyundertimepressure.Theneedforrapidswitchingisoneof
thereasonsthatAdd-3andmentalmultiplicationaresodifficult.ToperformtheAdd-3
task,youmustholdseveraldigitsinyourworkingmemoryatthesametime,associating
eachwithaparticularoperation:somedigitsareinthequeuetobetransformed,oneisin
theprocessoftransformation,andothers,alreadytransformed,areretainedforreporting.
Moderntestsofworkingmemoryrequiretheindividualtoswitchrepeatedlybetweentwo
demandingtasks,retainingtheresultsofoneoperationwhileperformingtheother.People
whodowellontheseteststendtodowellontestsofgeneralintelligence.However,the
abilitytocontrolattentionisnotsimplyameasureofintelligence;measuresofefficiency
inthecontrolofattentionpredictperformanceofairtrafficcontrollersandofIsraeliAir
Forcepilotsbeyondtheeffectsofintelligence.
Timepressureisanotherdriverofeffort.AsyoucarriedouttheAdd-3exercise,the
rushwasimposedinpartbythemetronomeandinpartbytheloadonmemory.Likea
juggler with several balls in the air, you cannot afford to slow down; the rate at which
material decays in memory forces the pace, driving you to refresh and rehearse
informationbeforeitislost.Anytaskthatrequiresyoutokeepseveralideasinmindatthe
same time has the same hurried character. Unless you have the good fortune of a
capacious working memory, you may be forced to work uncomfortably hard. The most
effortfulformsofslowthinkingarethosethatrequireyoutothinkfast.
YousurelyobservedasyouperformedAdd-3howunusualitisforyourmindtowork
sohard.Evenifyouthinkforaliving,fewofthementaltasksinwhichyouengageinthe
courseofaworkingdayareasdemandingasAdd-3,orevenasdemandingasstoringsix
digitsforimmediaterecall.Wenormallyavoidmentaloverloadbydividingourtasksinto
multiple easy steps, committing intermediate results to long-term memory or to paper
ratherthantoaneasilyoverloadedworkingmemory.Wecoverlongdistancesbytaking
ourtimeandconductourmentallivesbythelawofleasteffort.
SpeakingofAttentionandEffort
“I won’t try to solve this while driving. This is a pupil-dilating task. It requires
mentaleffort!”
“Thelawofleasteffortisoperatinghere.Hewillthinkaslittleaspossible.”
“She did not forget about the meeting. She was completely focused on something
elsewhenthemeetingwassetandshejustdidn’thearyou.”
“WhatcamequicklytomymindwasanintuitionfromSystem1.I’llhavetostart
overandsearchmymemorydeliberately.”
P
TheLazyController
IspendafewmonthseachyearinBerkeley,andoneofmygreatpleasuresthereisadaily
four-mile walk on a marked path in the hills, with a fine view of San Francisco Bay. I
usuallykeeptrackofmytimeandhavelearnedafairamountabouteffortfromdoingso.I
havefoundaspeed,about17minutesforamile,whichIexperienceasastroll.Icertainly
exertphysicaleffortandburnmorecaloriesatthatspeedthanifIsatinarecliner,butI
experiencenostrain,noconflict,andnoneedtopushmyself.Iamalsoabletothinkand
work while walking at that rate. Indeed, I suspect that the mild physical arousal of the
walkmayspilloverintogreatermentalalertness.
System 2 also has a natural speed. You expend some mental energy in random
thoughtsandinmonitoringwhatgoesonaroundyouevenwhenyourminddoesnothing
in particular, but there is little strain. Unless you are in a situation that makes you
unusuallywaryorself-conscious,monitoringwhathappensintheenvironmentorinside
yourhead demandslittleeffort.Youmake manysmalldecisions asyou driveyour car,
absorb some information as you read the newspaper, and conduct routine exchanges of
pleasantries with a spouse or a colleague, all with little effort and no strain. Just like a
stroll.
Itisnormallyeasyandactuallyquitepleasanttowalkandthinkatthesametime,but
attheextremestheseactivitiesappeartocompeteforthelimitedresourcesofSystem2.
Youcanconfirm this claimby a simpleexperiment.While walkingcomfortablywith a
friend,askhimtocompute23×78inhishead,andtodosoimmediately.Hewillalmost
certainlystopinhistracks.MyexperienceisthatIcanthinkwhilestrollingbutcannot
engage in mental work that imposes a heavy load on short-term memory. If I must
constructanintricateargumentundertimepressure,Iwouldratherbestill,andIwould
prefersitting to standing. Of course,not all slow thinking requiresthat formof intense
concentrationand effortfulcomputation—Idid thebest thinkingof my lifeon leisurely
walkswithAmos.
Accelerating beyond my strolling speed completely changes the experience of
walking,becausethetransitiontoafasterwalkbringsaboutasharpdeteriorationinmy
abilitytothinkcoherently.AsIspeedup,myattentionisdrawnwithincreasingfrequency
to the experience of walking and to the deliberate maintenance of the faster pace. My
abilitytobringatrainofthoughttoaconclusionisimpairedaccordingly.Atthehighest
speedIcansustainonthehills,about14minutesforamile,Idonoteventrytothinkof
anythingelse.Inadditiontothephysicaleffortofmovingmybodyrapidlyalongthepath,
amentaleffortofself-controlisneededtoresisttheurgetoslowdown.Self-controland
deliberatethoughtapparentlydrawonthesamelimitedbudgetofeffort.
Formostofus,mostofthetime,themaintenanceofacoherenttrainofthoughtand
theoccasionalengagementineffortfulthinkingalsorequireself-control.AlthoughIhave
notconductedasystematicsurvey,Isuspectthatfrequentswitchingoftasksandspeeded-
up mental work are not intrinsically pleasurable, and that people avoid them when
possible.Thisishowthelawofleasteffortcomestobealaw.Evenintheabsenceoftime
pressure,maintainingacoherenttrainofthoughtrequiresdiscipline.Anobserverofthe
numberoftimesIlookate-mailorinvestigatetherefrigeratorduringanhourofwriting
could wahene dd reasonably infer an urge to escape and conclude that keeping at it
requiresmoreself-controlthanIcanreadilymuster.
Fortunately, cognitive work is not always aversive, and people sometimes expend
considerable effort for long periods of time without having to exert willpower. The
psychologistMihalyCsikszentmihalyi(pronouncedsix-cent-mihaly)hasdonemorethan
anyone else to study this state of effortless attending, and the name he proposed for it,
flow,hasbecomepartofthelanguage.Peoplewhoexperienceflowdescribeitas“astate
of effortless concentration so deep that they lose their sense of time, of themselves, of
their problems,” and their descriptions of the joy of that state are so compelling that
Csikszentmihalyihascalleditan“optimalexperience.”Manyactivitiescaninduceasense
of flow, from painting to racing motorcycles—and for some fortunate authors I know,
evenwritingabookisoftenanoptimalexperience.Flowneatlyseparatesthetwoformsof
effort: concentration on the task and the deliberate control of attention. Riding a
motorcycleat150 miles anhourand playing acompetitivegameof chessarecertainly
very effortful. In a state of flow, however, maintaining focused attention on these
absorbing activities requires no exertion of self-control, thereby freeing resources to be
directedtothetaskathand.
TheBusyandDepletedSystem2
It is now a well-established proposition that both self-control and cognitive effort are
forms of mental work. Several psychological studies have shown that people who are
simultaneouslychallengedbyademandingcognitivetaskandbyatemptationaremore
likelytoyieldtothetemptation.Imaginethatyouareaskedtoretainalistofsevendigits
foraminuteortwo.Youaretoldthatrememberingthedigitsisyourtoppriority.While
yourattentionisfocusedonthedigits,youareofferedachoicebetweentwodesserts:a
sinfulchocolatecakeandavirtuousfruitsalad.Theevidencesuggeststhatyouwouldbe
morelikelytoselectthetemptingchocolatecakewhenyourmindisloadedwithdigits.
System1hasmoreinfluenceonbehaviorwhenSystem2isbusy,andithasasweettooth.
People who are cognitively busy are also more likely to make selfish choices, use
sexist language, and make superficial judgments in social situations. Memorizing and
repeatingdigitsloosenstheholdofSystem2onbehavior,butofcoursecognitiveloadis
nottheonlycauseofweakenedself-control.Afewdrinkshavethesameeffect,asdoesa
sleeplessnight.Theself-controlofmorningpeopleisimpairedatnight;thereverseistrue
of night people. Too much concern about how well one is doing in a task sometimes
disruptsperformancebyloadingshort-termmemorywithpointlessanxiousthoughts.The
conclusion is straightforward: self-control requires attention and effort. Another way of
saying this is that controlling thoughts and behaviors is one of the tasks that System 2
performs.
A series of surprising experiments by the psychologist Roy Baumeister and his
colleagues has shown conclusively that all variants of voluntary effort—cognitive,
emotional, or physical—draw at least partly on a shared pool of mental energy. Their
experimentsinvolvesuccessiveratherthansimultaneoustasks.
Baumeister’sgrouphasrepeatedlyfoundthataneffortofwillorself-controlistiring;
ifyouhavehadtoforceyourselftodosomething,youarelesswillingorlessabletoexert
self-controlwhenthenextchallengecomesaround.Thephenomenonhasbeennamedego
depletion.Inatypicaldemothypicaldenstration,participantswhoareinstructedtostifle
theiremotionalreactiontoanemotionallychargedfilmwilllaterperformpoorlyonatest
ofphysicalstamina—howlongtheycanmaintainastronggriponadynamometerinspite
ofincreasingdiscomfort.Theemotionaleffortinthefirstphaseoftheexperimentreduces
theabilitytowithstandthepainofsustainedmusclecontraction,andego-depletedpeople
thereforesuccumbmorequicklytotheurgetoquit.Inanotherexperiment,peoplearefirst
depleted by a task in which they eat virtuous foods such as radishes and celery while
resistingthetemptationtoindulgeinchocolateandrichcookies.Later,thesepeoplewill
giveupearlierthannormalwhenfacedwithadifficultcognitivetask.
Thelistofsituationsandtasksthatarenowknowntodepleteself-controlislongand
varied.Allinvolveconflictandtheneedtosuppressanaturaltendency.Theyinclude:
avoidingthethoughtofwhitebears
inhibitingtheemotionalresponsetoastirringfilm
makingaseriesofchoicesthatinvolveconflict
tryingtoimpressothers
respondingkindlytoapartner’sbadbehavior
interactingwithapersonofadifferentrace(forprejudicedindividuals)
Thelistofindicationsofdepletionisalsohighlydiverse:
deviatingfromone’sdiet
overspendingonimpulsivepurchases
reactingaggressivelytoprovocation
persistinglesstimeinahandgriptask
performingpoorlyincognitivetasksandlogicaldecisionmaking
Theevidenceispersuasive:activitiesthatimposehighdemandsonSystem2requireself-
control,andtheexertionofself-controlisdepletingandunpleasant.Unlikecognitiveload,
egodepletionisatleastinpartalossofmotivation.Afterexertingself-controlinonetask,
youdonotfeellikemakinganeffortinanother,althoughyoucoulddoitifyoureallyhad
to.In several experiments, people wereable to resist the effects of egodepletion when
givenastrongincentivetodoso.Incontrast,increasingeffortisnotanoptionwhenyou
mustkeepsixdigitsinshort-termmemorywhileperformingatask.Egodepletionisnot
thesamementalstateascognitivebusyness.
ThemostsurprisingdiscoverymadebyBaumeister’sgroupshows,asheputsit,that
theideaofmentalenergyismorethanameremetaphor.Thenervoussystemconsumes
moreglucosethanmostotherpartsofthebody,andeffortfulmentalactivityappearstobe
especially expensive in the currency of glucose. When you are actively involved in
difficult cognitive reasoning or engaged in a task that requires self-control, your blood
glucoseleveldrops.Theeffectisanalogoustoarunnerwhodrawsdownglucosestoredin
her muscles during a sprint. The bold implication of this idea is that the effects of ego
depletioncouldbeundonebyingestingglucose,andBaumeisterandhiscolleagueshave
confirmedthishypothesisnohypothesiinseveralexperiments.
Volunteers in one of their studies watched a short silent film of a woman being
interviewedandwereaskedtointerpretherbodylanguage.Whiletheywereperforming
thetask,aseriesofwordscrossed the screen inslowsuccession.Theparticipantswere
specificallyinstructedtoignorethewords,andiftheyfoundtheirattentiondrawnaway
theyhadtorefocustheirconcentrationonthewoman’sbehavior.Thisactofself-control
was known to cause ego depletion. All the volunteers drank some lemonade before
participatinginasecondtask.Thelemonadewassweetenedwithglucoseforhalfofthem
and with Splenda for the others. Then all participants were given a task in which they
needed to overcome an intuitive response to get the correct answer. Intuitive errors are
normally much more frequent among ego-depleted people, and the drinkers of Splenda
showedtheexpecteddepletioneffect.On theotherhand,theglucosedrinkers were not
depleted.Restoringthelevelofavailablesugarinthebrainhadpreventedthedeterioration
ofperformance.Itwilltakesometimeandmuchfurtherresearchtoestablishwhetherthe
tasks that cause glucose-depletion also cause the momentary arousal that is reflected in
increasesofpupilsizeandheartrate.
Adisturbingdemonstrationofdepletioneffectsinjudgmentwasrecentlyreportedin
theProceedings of the National Academy of Sciences.The unwittingparticipantsinthe
studywereeightparolejudgesinIsrael.Theyspendentiredaysreviewingapplicationsfor
parole.Thecasesarepresentedinrandomorder,andthejudgesspendlittletimeoneach
one, an average of 6 minutes. (The default decision is denial of parole; only 35% of
requestsareapproved.Theexacttimeofeachdecisionisrecorded,andthetimesofthe
judges’ three food breaks—morning break, lunch, and afternoon break—during the day
arerecordedaswell.)Theauthorsofthestudyplottedtheproportionofapprovedrequests
againstthe time sincethe lastfood break.The proportion spikesafter eachmeal, when
about 65% of requests are granted. During the two hours or so until the judges’ next
feeding,theapprovalratedropssteadily,toaboutzerojustbeforethemeal.Asyoumight
expect, this is an unwelcome result and the authors carefully checked many alternative
explanations.Thebestpossibleaccountofthedataprovidesbadnews:tiredandhungry
judgestendtofallbackontheeasierdefaultpositionofdenyingrequestsforparole.Both
fatigueandhungerprobablyplayarole.
TheLazySystem2
One of the main functions of System 2 is to monitor and control thoughts and actions
“suggested” by System 1, allowing some to be expressed directly in behavior and
suppressingormodifyingothers.
For an example, here is a simple puzzle. Do not try to solve it but listen to your
intuition:
Abatandballcost$1.10.
Thebatcostsonedollarmorethantheball.
Howmuchdoestheballcost?
Anumbercametoyourmind.Thenumber,ofcourse,is10:10¢.Thedistinctivemarkof
thiseasypuzzleisthatitevokesananswerthatisintuitive,appealing,andwrong.Dothe
math,andyouwillsee.Iftheballcosts10¢,thenthetotalcostwillbe$1.20(10¢forthe
balland$1.10forthebat),not$1.10.Thecorrectansweris5¢.It%”>5¢.issafetoassume
thattheintuitiveanswer also cametothemindofthosewhoendedupwiththecorrect
number—theysomehowmanagedtoresisttheintuition.
Shane Frederick and I worked together on a theory of judgment based on two
systems,andheusedthebat-and-ballpuzzletostudyacentralquestion:Howcloselydoes
System 2 monitor the suggestions of System 1? His reasoning was that we know a
significantfactaboutanyonewhosaysthattheballcosts10¢:thatpersondidnotactively
checkwhethertheanswerwascorrect,andherSystem2endorsedanintuitiveanswerthat
itcouldhaverejectedwithasmallinvestmentofeffort.Furthermore,wealsoknowthat
thepeoplewhogivetheintuitiveanswerhavemissedanobvioussocialcue;theyshould
have wondered why anyone would include in a questionnaire a puzzle with such an
obviousanswer.Afailuretocheckisremarkablebecausethecostofcheckingissolow:a
few seconds of mental work (the problem is moderately difficult), with slightly tensed
muscles and dilated pupils, could avoid an embarrassing mistake. People who say 10¢
appear to be ardent followers of the law of least effort. People who avoid that answer
appeartohavemoreactiveminds.
Manythousandsofuniversitystudentshaveansweredthebat-and-ballpuzzle,andthe
resultsareshocking.Morethan50%ofstudentsatHarvard,MIT,andPrincetontongave
theintuitive—incorrect—answer. Atless selectiveuniversities, therate ofdemonstrable
failure to check was in excess of 80%. The bat-and-ball problem is our first encounter
with an observation that will be a recurrent theme of this book: many people are
overconfident, prone to place too much faith in their intuitions. They apparently find
cognitiveeffortatleastmildlyunpleasantandavoiditasmuchaspossible.
Now I will show you a logical argument—two premises and a conclusion. Try to
determine,asquicklyasyoucan,iftheargumentislogicallyvalid.Doestheconclusion
followfromthepremises?
Allrosesareflowers.
Someflowersfadequickly.
Thereforesomerosesfadequickly.
Alargemajorityofcollegestudentsendorsethissyllogismasvalid.Infacttheargumentis
flawed,becauseitispossiblethattherearenorosesamongtheflowersthatfadequickly.
Just as in the bat-and-ball problem, a plausible answer comes to mind immediately.
Overriding it requires hard work—the insistent idea that “it’s true, it’s true!” makes it
difficulttocheckthelogic,andmostpeopledonottakethetroubletothinkthroughthe
problem.
This experiment has discouraging implications for reasoning in everyday life. It
suggeststhatwhenpeoplebelieveaconclusionistrue,theyarealsoverylikelytobelieve
argumentsthatappeartosupportit,evenwhentheseargumentsareunsound.IfSystem1
isinvolved,theconclusioncomesfirstandtheargumentsfollow.
Next,considerthefollowingquestionandansweritquicklybeforereadingon:
HowmanymurdersoccurinthestateofMichiganinoneyear?
Thequestion,whichwasalsodevisedbyShaneFrederick,isagainachallengetoSystem
2.The“trick”iswhethertherespondentwillrememberthatDetroit,ahigh-crimecthigh-
crimeity, is in Michigan. College students in the United States know this fact and will
correctlyidentifyDetroitasthelargestcityinMichigan.Butknowledgeofafactisnot
all-or-none.Factsthatweknowdonotalwayscometomindwhenweneedthem.People
whorememberthatDetroitisinMichigangivehigherestimatesofthemurderrateinthe
statethanpeoplewhodonot,butamajorityofFrederick’srespondentsdidnotthinkofthe
citywhenquestionedaboutthestate.Indeed,theaverageguessbypeoplewhowereasked
aboutMichiganislowerthanthe guesses ofasimilargroup whowereaskedabout the
murderrateinDetroit.
BlameforafailuretothinkofDetroitcanbelaidonbothSystem1andSystem2.
Whether the city comes to mind when the state is mentioned depends in part on the
automaticfunctionofmemory.Peopledifferinthisrespect.Therepresentationofthestate
ofMichiganisverydetailedinsomepeople’sminds:residentsofthestatearemorelikely
to retrieve many facts about it than people who live elsewhere; geography buffs will
retrievemorethanotherswhospecializeinbaseballstatistics;moreintelligentindividuals
aremorelikelythanotherstohaverichrepresentationsofmostthings.Intelligenceisnot
onlytheabilitytoreason;itisalsotheabilitytofindrelevantmaterialinmemoryandto
deploy attention when needed. Memory function is an attribute of System 1. However,
everyonehastheoptionofslowingdowntoconductanactivesearchofmemoryforall
possiblyrelevantfacts—justastheycouldslowdowntochecktheintuitiveanswerinthe
bat-and-ballproblem.Theextentofdeliberatecheckingandsearchisacharacteristicof
System2,whichvariesamongindividuals.
Thebat-and-ballproblem,theflowerssyllogism,andtheMichigan/Detroitproblem
havesomethingincommon.Failingtheseminitestsappearstobe,atleasttosomeextent,
amatterofinsufficientmotivation,nottryinghardenough.Anyonewhocanbeadmitted
to a good university is certainly able to reason through the first two questions and to
reflectaboutMichiganlongenoughtorememberthemajorcityinthatstateanditscrime
problem. These students can solve much more difficult problems when they are not
temptedtoacceptasuperficiallyplausibleanswerthatcomesreadilytomind.Theease
withwhichtheyaresatisfiedenoughtostopthinkingisrathertroubling.“Lazy”isaharsh
judgmentabouttheself-monitoringoftheseyoungpeopleandtheirSystem2,butitdoes
not seem to be unfair. Those who avoid the sin of intellectual sloth could be called
“engaged.”Theyaremorealert,moreintellectuallyactive,lesswillingtobesatisfiedwith
superficially attractive answers, more skeptical about their intuitions. The psychologist
KeithStanovichwouldcallthemmorerational.
Intelligence,Control,Rationality
Researchers have applied diverse methods to examine the connection between thinking
andself-control.Somehaveaddresseditbyaskingthecorrelationquestion:Ifpeoplewere
rankedbytheirself-controlandbytheircognitiveaptitude,wouldindividualshavesimilar
positionsinthetworankings?
Inoneofthemostfamousexperimentsinthehistoryofpsychology,WalterMischel
and his students exposed four-year-old children to a cruel dilemma. They were given a
choicebetweenasmallreward(oneOreo),whichtheycouldhaveatanytime,oralarger
reward (two cookies) for which they had to wait 15 minutes under difficult conditions.
Theyweretoremainaloneinaroom,facingadeskwithtwoobjects:asinglecookieanda
bellthat thechild could ring at anytime to call in the experimenter andreceiven oand
recei the one cookie. As the experiment was described: “There were no toys, books,
pictures,orotherpotentiallydistractingitemsintheroom.Theexperimenterlefttheroom
and did not return until 15 min had passed or the child had rung the bell, eaten the
rewards,stoodup,orshownanysignsofdistress.”
Thechildrenwerewatchedthroughaone-waymirror,andthefilmthatshowstheir
behaviorduringthewaitingtimealwayshastheaudienceroaringinlaughter.Abouthalf
thechildrenmanagedthefeatofwaitingfor15minutes,mainlybykeepingtheirattention
awayfromthetemptingreward.Tenorfifteenyearslater,alargegaphadopenedbetween
those who had resisted temptation and those who had not. The resisters had higher
measuresof executivecontrolin cognitivetasks, andespeciallythe abilityto reallocate
their attention effectively. As young adults, they were less likely to take drugs. A
significantdifferenceinintellectualaptitudeemerged:thechildrenwhohadshownmore
self-controlasfour-year-oldshadsubstantiallyhigherscoresontestsofintelligence.
AteamofresearchersattheUniversityofOregonexploredthelinkbetweencognitive
control and intelligence in several ways, including an attempt to raise intelligence by
improvingthecontrolofattention.Duringfive40-minutesessions,theyexposedchildren
agedfourtosixtovariouscomputergamesespeciallydesignedtodemandattentionand
control. In one of the exercises, the children used a joystick to track a cartoon cat and
moveittoagrassyareawhileavoidingamuddyarea.Thegrassyareasgraduallyshrank
andthemuddyareaexpanded,requiringprogressivelymoreprecisecontrol.Thetesters
found that training attention not only improved executive control; scores on nonverbal
tests of intelligence also improved and the improvement was maintained for several
months.Otherresearchbythesamegroupidentifiedspecificgenesthatareinvolvedinthe
control of attention, showed that parenting techniques also affected this ability, and
demonstratedacloseconnectionbetweenthechildren’sabilitytocontroltheir attention
andtheirabilitytocontroltheiremotions.
ShaneFrederickconstructedaCognitiveReflectionTest,whichconsistsofthebat-
and-ball problem and two other questions, chosen because they also invite an intuitive
answerthatisbothcompellingandwrong(thequestionsareshownhere).Hewentonto
study the characteristics of students who score very low on this test—the supervisory
functionofSystem2isweakinthesepeople—andfoundthattheyarepronetoanswer
questionswiththefirstideathatcomestomindandunwillingtoinvesttheeffortneeded
tochecktheirintuitions.Individualswhouncriticallyfollowtheirintuitionsaboutpuzzles
arealsopronetoacceptothersuggestionsfromSystem1.Inparticular,theyareimpulsive,
impatient,andkeentoreceiveimmediategratification.Forexample,63%oftheintuitive
respondentssaytheywouldprefertoget$3,400thismonthratherthan$3,800nextmonth.
Only 37% of those who solve all three puzzles correctly have the same shortsighted
preferenceforreceivingasmalleramountimmediately.Whenaskedhowmuchtheywill
pay to get overnight delivery of a book they have ordered, the low scorers on the
CognitiveReflectionTestarewillingtopaytwiceasmuchasthehighscorers.Frederick’s
findings suggest that the characters of our psychodrama have different “personalities.”
System1isimpulsiveandintuitive;System2iscapableofreasoning,anditiscautious,
but at least for some people it is also lazy. We recognize related differences among
individuals:somepeoplearemoreliketheirSystem2;othersareclosertotheirSystem1.
Thissimpletesthasemergedasoneofthebetterpredictorsoflaztestorsoflythinking.
KeithStanovichandhislongtimecollaboratorRichardWestoriginallyintroducedthe
termsSystem1andSystem2(theynowprefertospeakofType1andType2processes).
Stanovichandhiscolleagueshavespentdecadesstudyingdifferencesamongindividuals
inthekindsofproblemswithwhichthisbookisconcerned.Theyhaveaskedonebasic
questioninmanydifferentways:Whatmakessomepeoplemoresusceptiblethanothersto
biasesofjudgment?StanovichpublishedhisconclusionsinabooktitledRationalityand
the Reflective Mind, which offers a bold and distinctive approach to the topic of this
chapter. He draws a sharp distinction between two parts of System 2—indeed, the
distinctionissosharpthathecallsthemseparate“minds.”Oneoftheseminds(hecallsit
algorithmic)dealswithslowthinkinganddemandingcomputation.Somepeoplearebetter
than others in these tasks of brain power—they are the individuals who excel in
intelligencetestsandareabletoswitchfromonetasktoanotherquicklyandefficiently.
However,Stanovicharguesthathighintelligencedoesnotmakepeopleimmunetobiases.
Anotherabilityisinvolved,whichhelabelsrationality.Stanovich’sconceptofarational
person is similar to what I earlier labeled “engaged.” The core of his argument is that
rationality should be distinguished from intelligence. In his view, superficial or “lazy”
thinkingisaflawinthereflectivemind,afailureofrationality.Thisisanattractiveand
thought-provokingidea.Insupportofit,Stanovichandhiscolleagueshavefoundthatthe
bat-and-ballquestionandotherslikeitaresomewhatbetterindicatorsofoursusceptibility
tocognitiveerrorsthanareconventionalmeasuresofintelligence,suchasIQtests.Time
will tell whether the distinction between intelligence and rationality can lead to new
discoveries.
SpeakingofControl
“Shedidnothavetostruggletostayontaskforhours.Shewasinastateofflow.”
“His ego was depleted after a long day of meetings. So he just turned to standard
operatingproceduresinsteadofthinkingthroughtheproblem.”
“Hedidn’tbothertocheckwhetherwhathesaidmadesense.Doesheusuallyhavea
lazySystem2orwasheunusuallytired?”
“Unfortunately, she tends to say the first thing that comes into her mind. She
probablyalsohastroubledelayinggratification.WeakSystem2.”
P
TheAssociativeMachine
TobeginyourexplorationofthesurprisingworkingsofSystem1,lookatthefollowing
words:
BananasVomit
Alothappenedtoyouduringthelastsecondortwo.Youexperiencedsomeunpleasant
imagesandmemories.Yourfacetwistedslightlyinanexpressionofdisgust,andyoumay
havepushedthisbookimperceptiblyfartheraway.Yourheartrateincreased,thehairon
yourarmsrosealittle,andyoursweatglandswereactivated.Inshort,yourespondedto
thedisgustingwordwithanattenuatedversionofhowyouwouldreacttotheactualevent.
Allofthiswascompletelyautomatic,beyondyourcontrol.
There was no particular reason to do so, but your mind automatically assumed a
temporal sequence and a causal connection between the words bananas and vomit,
forming a sketchy scenario in which bananas caused the sickness. As a result, you are
experiencingatemporaryaversiontobananas(don’tworry,itwillpass).Thestateofyour
memory has changed in other ways: you are now unusually ready to recognize and
respondtoobjectsandconcepts associated with“vomit,”suchassick,stink,ornausea,
and words associated with “bananas,” such as yellow and fruit, and perhaps apple and
berries.
Vomiting normally occurs in specific contexts, such as hangovers and indigestion.
Youwouldalsobeunusuallyreadytorecognizewordsassociatedwithothercausesofthe
same unfortunate outcome. Furthermore, your System 1 noticed the fact that the
juxtapositionofthetwowordsisuncommon;youprobablyneverencountereditbefore.
Youexperiencedmildsurprise.
This complex constellation of responses occurred quickly, automatically, and
effortlessly.Youdidnotwillitandyoucouldnotstopit.ItwasanoperationofSystem1.
The events that took place as a result of your seeing the words happened by a process
calledassociativeactivation:ideasthathavebeenevokedtriggermanyotherideas,ina
spreadingcascadeofactivityinyourbrain.Theessentialfeatureof this complexsetof
mental events is its coherence. Each element is connected, and each supports and
strengthenstheothers.Thewordevokesmemories,whichevokeemotions,whichinturn
evoke facial expressions and other reactions, such as a general tensing up and an
avoidancetendency.Thefacialexpressionandtheavoidancemotionintensifythefeelings
to which they are linked, and the feelings in turn reinforce compatible ideas. All this
happensquicklyandallatonce,yieldingaself-reinforcingpatternofcognitive,emotional,
andphysicalresponsesthatisbothdiverseandintegrated—ithasbeencalledassociatively
coherent.
Inasecondorsoyouaccomplished,automaticallyandunconsciously,aremarkable
feat.Startingfromacompletelyunexpectedevent,yourSystem1madeasmuchsenseas
possibleofthesituation—twosimplewords,oddlyjuxtaposed—bylinkingthewordsina
causalstory;itevaluatedthepossiblethreat(mildtomoderate)andcreatedacontextfor
futuredevelopmentsbypreparingyouforeventsthathadjustbecomemorelikely;italso
createdacontextforthecurrenteventbyevaluatinghowsurprisingitwas.Youendedup
asinformedaboutthepastandaspreparedforthefutureasyoucouldbe.
AnoddfeatureofwhathappenedisthatyourSystem1treatedthemereconjunction
oftwowordsasrepresentationsofreality.Yourbodyreactedinanattenuatedreplicaofa
reactiontotherealthing,andtheemotionalresponseandphysicalrecoilwerepartofthe
interpretation of the event. As cognitive scientists have emphasized in recent years,
cognitionisembodied;youthinkwithyourbody,notonlywithyourbrain.
Themechanismthatcausesthesementaleventshasbeenknownforalongtime:itis
theass12;velyociationofideas.Weallunderstandfromexperiencethatideasfolloweach
other in our conscious mind in a fairly orderly way. The British philosophers of the
seventeenthandeighteenthcenturiessearchedfortherulesthatexplainsuchsequences.In
An Enquiry Concerning Human Understanding, published in 1748, the Scottish
philosopher David Hume reduced the principles of association to three: resemblance,
contiguity in time and place, and causality. Our concept of association has changed
radicallysinceHume’sdays,buthisthreeprinciplesstillprovideagoodstart.
Iwilladoptanexpansiveviewofwhatanideais.Itcanbeconcreteorabstract,andit
canbeexpressedinmanyways:asaverb,asanoun,asanadjective,orasaclenchedfist.
Psychologists think of ideas as nodes in a vast network, called associative memory, in
whicheachideais linked tomanyothers.Thereare differenttypesoflinks:causes are
linkedtotheireffects(virus cold);thingstotheirproperties(lime green);thingsto
thecategoriestowhichtheybelong(banana fruit).Onewaywehaveadvancedbeyond
Humeisthatwenolongerthinkofthemindasgoingthroughasequenceofconscious
ideas,oneatatime.Inthecurrentviewofhowassociativememoryworks,agreatdeal
happensatonce.Anideathathasbeenactivateddoesnotmerelyevokeoneotheridea.It
activates many ideas, which in turn activate others. Furthermore, only a few of the
activatedideaswillregisterinconsciousness;mostoftheworkofassociativethinkingis
silent,hidden from our conscious selves.The notionthat we havelimited accessto the
workingsofourmindsisdifficulttoacceptbecause,naturally,itisalientoourexperience,
butitistrue:youknowfarlessaboutyourselfthanyoufeelyoudo.
TheMarvelsofPriming
As is common in science, the first big breakthrough in our understanding of the
mechanismofassociationwasanimprovementinamethodofmeasurement.Untilafew
decadesago,theonlywaytostudyassociationswastoaskmanypeoplequestionssuchas,
“What is the first word that comes to your mind when you hear the word DAY?” The
researcherstalliedthefrequencyofresponses,suchas“night,”“sunny,”or“long.”Inthe
1980s, psychologists discovered that exposure to a word causes immediate and
measurable changes in the ease with which many related words can be evoked. If you
haverecentlyseenorheardthewordEAT,youaretemporarilymorelikelytocompletethe
wordfragmentSO_PasSOUPthanasSOAP.Theoppositewouldhappen,ofcourse,if
youhadjustseenWASH.WecallthisaprimingeffectandsaythattheideaofEATprimes
theideaofSOUP,andthatWASHprimesSOAP.
Priming effects take many forms. If the idea of EAT is currently on your mind
(whetherornotyouareconsciousofit),youwillbequickerthanusualtorecognizethe
wordSOUPwhenitisspokeninawhisperorpresentedinablurryfont.Andofcourse
youareprimednotonlyfortheideaofsoupbutalsoforamultitudeoffood-relatedideas,
including fork, hungry, fat, diet, and cookie. If for your most recent meal you sat at a
wobblyrestauranttable,youwillbeprimedforwobblyaswell.Furthermore,theprimed
ideas have some ability to prime other ideas, although more weakly. Like ripples on a
pond,activationspreadsthroughasmallpartofthevastnetworkofassociatedideas.The
mapping of these ripples is now one of the most exciting pursuits in psychological
research.
Another major advance in our understanding of memory was the discovery that
priming is not restricted to concepts and words. You cannot know this from conscious
experience, of course, but you must accept the alien idea that your actions and your
emotionscanbeprimedbyeventsofwhichyouarenotevenaware.Inanexperimentthat
became an instant classic, the psychologist John Bargh and his collaborators asked
studentsatNewYorkUniversity—mostagedeighteentotwenty-two—toassemblefour-
wordsentencesfromasetoffivewords(forexample,“findsheityellowinstantly”).For
onegroupofstudents,halfthescrambledsentencescontainedwordsassociatedwiththe
elderly,suchasFlorida,forgetful,bald,gray,orwrinkle.Whentheyhadcompletedthat
task,theyoungparticipantsweresentouttodoanotherexperimentinanofficedownthe
hall.Thatshortwalkwaswhattheexperimentwasabout.Theresearchersunobtrusively
measuredthetimeittookpeopletogetfromoneendofthecorridortotheother.AsBargh
hadpredicted,theyoungpeoplewhohadfashionedasentencefromwordswithanelderly
themewalkeddownthehallwaysignificantlymoreslowlythantheothers.
The “Florida effect” involves two stages of priming. First, the set of words primes
thoughtsofoldage,thoughthewordoldisnevermentioned;second,thesethoughtsprime
abehavior,walkingslowly,whichisassociatedwitholdage.Allthishappenswithoutany
awareness.Whentheywerequestionedafterward,noneofthestudentsreportednoticing
thatthewordshadhadacommontheme,andtheyallinsistedthatnothingtheydidafter
thefirstexperimentcouldhavebeeninfluencedbythewordstheyhadencountered.The
ideaofoldagehadnotcometotheirconsciousawareness,buttheiractionshadchanged
nevertheless.Thisremarkableprimingphenomenon—theinfluencingofanactionbythe
idea—isknownastheideomotoreffect.Althoughyousurelywerenotawareofit,reading
thisparagraphprimedyouaswell.Ifyouhadneededtostanduptogetaglassofwater,
you would have been slightly slower than usual to rise from your chair—unless you
happentodisliketheelderly,inwhichcaseresearchsuggeststhatyoumighthavebeen
slightlyfasterthanusual!
Theideomotorlinkalsoworksinreverse.AstudyconductedinaGermanuniversity
wasthemirrorimageoftheearlyexperimentthatBarghandhiscolleagueshadcarriedout
inNewYork.Studentswereaskedtowalkaroundaroomfor5minutesatarateof30
stepsperminute,whichwasaboutone-thirdtheirnormalpace.Afterthisbriefexperience,
the participants were much quicker to recognize words related to old age, such as
forgetful,old,andlonely.Reciprocalprimingeffectstendtoproduceacoherentreaction:
ifyouwereprimedtothinkofoldage,youwouldtendtoactold,andactingoldwould
reinforcethethoughtofoldage.
Reciprocallinksarecommonintheassociativenetwork.Forexample,beingamused
tendstomakeyousmile,andsmilingtendstomakeyoufeelamused.Goaheadandtakea
pencil,andholditbetweenyourteethforafewsecondswiththeeraserpointingtoyour
rightandthepointtoyourleft.Nowholdthepencilsothepointisaimedstraightinfront
ofyou,bypursingyourlipsaroundtheeraserend.Youwereprobablyunawarethatoneof
theseactionsforcedyourfaceintoafrownandtheotherintoasmile.Collegestudents
wereaskedtoratethehumorofcartoonsfromGaryLarson’sTheFarSidewhileholding
apencilintheirmouth.Thosewhowere“smiling”(withoutanyawarenessofdoingso)
foundthecartoonsrri221;(withfunnierthandidthosewhowere“frowning.”Inanother
experiment, people whose face was shaped into a frown (by squeezing their eyebrows
together) reported an enhanced emotional response to upsetting pictures—starving
children,peoplearguing,maimedaccidentvictims.
Simple,commongesturescanalsounconsciouslyinfluenceourthoughtsandfeelings.
Inonedemonstration,peoplewereaskedtolistentomessagesthroughnewheadphones.
They were told that the purpose of the experiment was to test the quality of the audio
equipmentandwereinstructedtomovetheirheadsrepeatedlytocheckforanydistortions
ofsound.Halftheparticipantsweretoldtonodtheirheadupanddownwhileotherswere
toldtoshakeitsidetoside.Themessagestheyheardwereradioeditorials.Thosewho
nodded(ayesgesture)tendedtoacceptthemessagetheyheard,butthosewhoshooktheir
headtendedtorejectit.Again,therewasnoawareness,justahabitualconnectionbetween
anattitudeofrejectionoracceptance and itscommonphysicalexpression.Youcansee
whythecommonadmonitionto“actcalmandkindregardlessofhowyoufeel”isvery
goodadvice:youarelikelytoberewardedbyactuallyfeelingcalmandkind.
PrimesThatGuideUs
Studies of priming effects have yielded discoveries that threaten our self-image as
consciousandautonomousauthorsofourjudgmentsandourchoices.Forinstance,most
ofus thinkof voting as a deliberateact thatreflectsour valuesand ourassessments of
policies and is not influenced by irrelevancies. Our vote should not be affected by the
locationofthepollingstation,forexample,butitis.Astudyofvotingpatternsinprecincts
of Arizona in 2000 showed that the support for propositions to increase the funding of
schoolswassignificantlygreaterwhenthepollingstationwasinaschoolthanwhenitwas
in a nearby location. A separate experiment showed that exposing people to images of
classrooms and school lockers also increased the tendency of participants to support a
schoolinitiative.Theeffectoftheimageswaslargerthanthedifferencebetweenparents
and other voters! The study of priming has come some way from the initial
demonstrationsthatremindingpeopleofoldagemakesthemwalkmoreslowly.Wenow
knowthattheeffectsofprimingcanreachintoeverycornerofourlives.
Remindersofmoneyproducesometroublingeffects.Participantsinoneexperiment
wereshownalistoffivewordsfromwhichtheywererequiredtoconstructafour-word
phrase that had a money theme (“high a salary desk paying” became “a high-paying
salary”). Other primes were much more subtle, including the presence of an irrelevant
money-relatedobjectinthebackground,suchasastackofMonopolymoneyonatable,or
acomputerwithascreensaverofdollarbillsfloatinginwater.
Money-primed people become more independent than they would be without the
associativetrigger.Theyperseveredalmosttwiceaslongintryingtosolveaverydifficult
problembeforetheyaskedtheexperimenterforhelp,acrispdemonstrationofincreased
self-reliance.Money-primedpeoplearealsomoreselfish:theyweremuchlesswillingto
spendtimehelpinganotherstudentwhopretendedtobeconfusedaboutanexperimental
task. When an experimenter clumsily dropped a bunch of pencils on the floor, the
participantswithmoney(unconsciously)ontheirmindpickedupfewerpencils.Inanother
experiment in the series, participants were told that they would shortly have a get-
acquaintedconversationwithanotherpersonandwereaskedtosetuptwochairswhilethe
experimenterlefttoretrievethatperson.Participantsprimedbymoneychoseintheexto
stay much farther apart than their nonprimed peers (118 vs. 80 centimeters). Money-
primedundergraduatesalsoshowedagreaterpreferenceforbeingalone.
Thegeneralthemeofthesefindingsisthattheideaofmoneyprimesindividualism:a
reluctance to be involved with others, to depend on others, or to accept demands from
others.Thepsychologistwhohasdonethisremarkableresearch,KathleenVohs,hasbeen
laudablyrestrainedindiscussingtheimplicationsofherfindings,leavingthetasktoher
readers.Herexperimentsareprofound—herfindingssuggestthatlivinginaculturethat
surroundsuswithremindersofmoneymayshapeourbehaviorandourattitudesinways
thatwe do not know aboutand of which we maynot be proud. Some culturesprovide
frequentremindersofrespect,othersconstantlyremindtheirmembersofGod,andsome
societiesprimeobediencebylargeimagesoftheDearLeader.Cantherebeanydoubtthat
theubiquitousportraitsofthenationalleaderindictatorialsocietiesnotonlyconveythe
feelingthat“BigBrotherIsWatching”butalsoleadtoanactualreductioninspontaneous
thoughtandindependentaction?
The evidence of priming studies suggests that reminding people of their mortality
increasestheappealofauthoritarianideas,whichmaybecomereassuringinthecontextof
theterrorofdeath.OtherexperimentshaveconfirmedFreudianinsightsabouttheroleof
symbolsandmetaphorsinunconsciousassociations.Forexample,considertheambiguous
wordfragmentsW__HandS__P.Peoplewhowererecentlyaskedtothinkofanaction
of which they are ashamed are more likely to complete those fragments as WASH and
SOAP and less likely to see WISH and SOUP. Furthermore, merely thinking about
stabbingacoworkerinthebackleavespeoplemoreinclinedtobuysoap,disinfectant,or
detergentthanbatteries,juice,orcandybars.Feelingthatone’ssoulisstainedappearsto
triggeradesiretocleanseone’sbody,animpulsethathasbeendubbedthe“LadyMacbeth
effect.”
Thecleansingishighlyspecifictothebodypartsinvolvedinasin.Participantsinan
experimentwereinducedto“lie”toanimaginaryperson,eitheronthephoneorine-mail.
Ina subsequenttest ofthe desirabilityof variousproducts, peoplewho hadlied onthe
phonepreferredmouthwashoversoap,andthosewhohadliedine-mailpreferredsoapto
mouthwash.
WhenIdescribeprimingstudiestoaudiences,thereactionisoftendisbelief.Thisis
notasurprise:System2believesthatitisinchargeandthatitknowsthereasonsforits
choices.Questionsareprobablycroppingupinyourmindaswell:Howisitpossiblefor
suchtrivialmanipulationsofthecontexttohavesuchlargeeffects?Dotheseexperiments
demonstrate that we are completely at the mercy of whatever primes the environment
provides at any moment? Of course not. The effects of the primes are robust but not
necessarily large. Among a hundred voters, only a few whose initial preferences were
uncertainwillvotedifferentlyaboutaschoolissueiftheirprecinctislocatedinaschool
ratherthaninachurch—butafewpercentcouldtipanelection.
Theideayoushouldfocuson,however,isthatdisbeliefisnotanoption.Theresults
arenotmadeup,noraretheystatisticalflukes.Youhavenochoicebuttoacceptthatthe
majorconclusionsofthesestudiesaretrue.Moreimportant,youmustacceptthattheyare
trueaboutyou.Ifyouhadbeenexposedtoascreensaveroffloatingdollarbills,youtoo
wouldlikelyhavepickedupfewerpencilstohelpaclumsystranger.Youdonotbelieve
that these results apply to you because they correspond to nothing in your subjective
experience.Butyoursubjectiveexpefteelief.Trienceconsistslargelyofthestorythatyour
System2tellsitselfaboutwhatisgoingon.PrimingphenomenaariseinSystem1,and
youhavenoconsciousaccesstothem.
Iconcludewithaperfectdemonstrationofaprimingeffect,whichwasconductedin
anofficekitchenataBritishuniversity.Formanyyearsmembersofthatofficehadpaid
fortheteaorcoffeetowhichtheyhelpedthemselvesduringthedaybydroppingmoney
intoan“honestybox.”Alistofsuggestedpriceswasposted.Onedayabannerposterwas
displayed just above the price list, with no warning or explanation. For aperiod of ten
weeksanewimagewaspresentedeachweek,eitherflowersoreyesthatappearedtobe
looking directly at the observer. No one commented on the new decorations, but the
contributionstothehonestyboxchangedsignificantly.Thepostersandtheamountsthat
peopleputintothecashbox(relativetotheamounttheyconsumed)areshowninfigure4.
Theydeserveacloselook.
Figure4
Onthefirstweekoftheexperiment(whichyoucanseeatthebottomofthefigure),two
wide-open eyes stare at the coffee or tea drinkers, whose average contribution was 70
penceperliterofmilk.Onweek2,thepostershowsflowersandaveragecontributions
drop to about 15 pence. The trend continues. On average, the users of the kitchen
contributed almost three times as much in “eye weeks” as they did in “flowerweeks.”
Evidently, a purely symbolic reminder of being watched prodded people into improved
behavior.Asweexpectatthispoint,theeffectoccurswithoutanyawareness.Doyounow
believethatyouwouldalsofallintothesamepattern?
Someyearsago,thepsychologistTimothyWilsonwroteabookwiththeevocative
titleStrangerstoOurselves.Youhavenowbeenintroducedtothatstrangerinyou,which
may be in control of much of what you do, although you rarely have a glimpse of it.
System1providestheimpressionsthatoftenturnintoyourbeliefs,andisthesourceofthe
impulsesthatoftenbecomeyourchoicesandyouractions.Itoffersatacitinterpretationof
whathappens toyou andaround you, linking the present withthe recentpast and with
expectations about the near future. It contains the model of the world that instantly
evaluateseventsasnormalorsurprising.Itisthesourceofyourrapidandoftenprecise
intuitive judgments. And it does most of this without your conscious awareness of its
activities.System1isalso,aswewillseeinthefollowingchapters,theoriginofmanyof
thesystematicerrorsinyourintuitions.
SpeakingofPriming
“Thesightofallthesepeopleinuniformsdoesnotprimecreativity.”
“The world makes much less sense than you think. The coherence comes mostly
fromthewayyourmindworks.”
“Theywereprimedtofindflaws,andthisisexactlywhattheyfound.”
“HisSystem1constructedastory,andhisSystem2believedit.Ithappenstoallel
“ImademyselfsmileandI’mactuallyfeelingbetter!”
P
CognitiveEase
Wheneveryouareconscious,andperhapsevenwhenyouarenot,multiplecomputations
are going on in your brain, which maintain and update current answers to some key
questions:Isanythingnewgoingon?Isthereathreat?Arethingsgoingwell?Shouldmy
attentionberedirected?Ismoreeffortneededforthistask?Youcanthinkofacockpit,
withasetofdialsthatindicatethecurrentvaluesofeachoftheseessentialvariables.The
assessments are carried out automatically by System 1, and one of their functions is to
determinewhetherextraeffortisrequiredfromSystem2.
One of the dials measures cognitive ease, and its range is between “Easy” and
“Strained.”Easyisasignthatthingsaregoingwell—nothreats,nomajornews,noneed
toredirectattentionormobilizeeffort.Strainedindicatesthataproblemexists,whichwill
requireincreasedmobilizationofSystem2.Conversely,youexperiencecognitivestrain.
Cognitivestrainisaffectedbyboththecurrentlevelofeffortandthepresenceofunmet
demands. The surprise is that a single dial of cognitive ease is connected to a large
networkofdiverseinputsandoutputs.Figure5tellsthestory.
Thefiguresuggeststhatasentencethatisprintedinaclearfont,orhasbeenrepeated,
or has been primed, will be fluently processed with cognitive ease. Hearing a speaker
whenyouareinagoodmood,orevenwhenyouhaveapencilstuckcrosswiseinyour
mouth to make you “smile,” also induces cognitive ease. Conversely, you experience
cognitivestrainwhenyoureadinstructionsinapoorfont,orinfaintcolors,orwordedin
complicatedlanguage,orwhenyouareinabadmood,andevenwhenyoufrown.
Figure5.CausesandConsequencesofCognitiveEase
Thevariouscausesofeaseorstrainhaveinterchangeableeffects.Whenyouareina
stateofcognitiveease,youareprobablyinagoodmood,likewhatyousee,believewhat
youhear,trustyourintuitions,andfeelthatthecurrentsituationiscomfortablyfamiliar.
Youarealsolikelytoberelativelycasualandsuperficialinyourthinking.Whenyoufeel
strained,youaremorelikelytobevigilantandsuspicious,investmoreeffortinwhatyou
aredoing,feellesscomfortable,andmakefewererrors,butyoualsoarelessintuitiveand
lesscreativethanusual.
IllusionsofRemembering
Thewordillusionbringsvisualillusionstomind,becauseweareallfamiliarwithpictures
thatmislead.Butvisionisnottheonlydomainofillusions;memoryisalsosusceptibleto
them,asisthinkingmoregenerally.
DavidStenbill,MonicaBigoutski,Sh”imight=sispictanaTirana.Ijustmadeupthese
names. If you encounter any of them within the next few minutes you are likely to
rememberwhereyousawthem.Youknow,andwillknowforawhile,thatthesearenot
thenamesofminorcelebrities.Butsupposethatafewdaysfromnowyouareshowna
longlistofnames,includingsomeminorcelebritiesand“new”namesofpeoplethatyou
haveneverheardof;yourtaskwillbetocheckeverynameofacelebrityinthelist.There
isasubstantialprobabilitythatyouwillidentifyDavidStenbillasawell-knownperson,
althoughyouwillnot(ofcourse)knowwhetheryouencounteredhisnameinthecontext
ofmovies,sports,orpolitics.LarryJacoby,thepsychologistwhofirstdemonstratedthis
memoryillusioninthelaboratory,titledhisarticle“BecomingFamousOvernight.”How
does this happen? Start by asking yourself how you know whether or not someone is
famous.Insomecasesoftrulyfamouspeople(orofcelebritiesinanareayoufollow),you
have a mental file with rich information about a person—think Albert Einstein, Bono,
Hillary Clinton. But you will have no file of information about David Stenbill if you
encounterhisnameinafewdays.Allyouwillhaveisasenseoffamiliarity—youhave
seenthisnamesomewhere.
Jacoby nicely stated the problem: “The experience of familiarity has a simple but
powerfulqualityof‘pastness’thatseemstoindicatethatitisadirectreflectionofprior
experience.” This quality of pastness is an illusion. The truth is, as Jacoby and many
followers have shown, that the name David Stenbill will look familiar when you see it
becauseyouwillseeitmoreclearly.Wordsthatyouhaveseenbeforebecomeeasiertosee
again—youcanidentifythembetterthanotherwordswhentheyareshownverybrieflyor
maskedbynoise,andyouwillbequicker(byafewhundredthsofasecond)toreadthem
thantoreadotherwords.Inshort,youexperiencegreatercognitiveeaseinperceivinga
wordyouhaveseenearlier,anditisthissenseofeasethatgivesyoutheimpressionof
familiarity.
Figure5suggestsawaytotestthis.Chooseacompletelynewword,makeiteasierto
see,anditwillbemorelikelytohavethequalityofpastness.Indeed,anewwordismore
likelytoberecognizedasfamiliarifitisunconsciouslyprimedbyshowingitforafew
milliseconds just before the test, or if it is shown in sharper contrast than some other
wordsinthelist.Thelinkalsooperatesintheotherdirection.Imagineyouareshownalist
ofwordsthataremoreorlessoutoffocus.Someofthewordsareseverelyblurred,others
lessso,andyourtaskistoidentifythewordsthatareshownmoreclearly.Awordthatyou
haveseenrecentlywillappeartobeclearerthanunfamiliarwords.Asfigure5indicates,
the various ways of inducing cognitive ease or strain are interchangeable; you may not
knowpreciselywhatitisthatmakesthingscognitivelyeasyorstrained.Thisishowthe
illusionoffamiliaritycomesabout.
IllusionsofTruth
“NewYorkisalargecityintheUnitedStates.”“ThemoonrevolvesaroundEarth.”“A
chicken has four legs.” In all these cases, you quickly retrieved a great deal of related
information,almostallpointingonewayoranother.Youknewsoonafterreadingthem
that the first two statements are true and the last one is false. Note, however, that the
statement “A chicken has three legs” is more obviously false than “A chicken has four
legs.”Yourassociativemachineryslowsthejudgmentofthelattersentencebydelivering
thefactthatmanyanimalshavefourlegs,andperhapsalsothatsupermarketsoftensell
chickenordblurred, legs in packages of four. System 2 was involved in sifting that
information,perhaps raising theissueofwhetherthequestionaboutNewYorkwastoo
easy,orcheckingthemeaningofrevolves.
Thinkofthelasttimeyoutookadrivingtest.Isittruethatyouneedaspeciallicense
todriveavehiclethatweighsmorethanthreetons?Perhapsyoustudiedseriouslyandcan
rememberthesideofthepageonwhichtheanswerappeared,aswellasthelogicbehind
it. This is certainly not how I passed driving tests when I moved to a new state. My
practicewastoreadthebookletofrulesquicklyonceandhopeforthebest.Iknewsome
oftheanswersfromtheexperienceofdrivingforalongtime.Buttherewerequestions
wherenogoodanswercametomind,whereallIhadtogobywascognitiveease.Ifthe
answerfeltfamiliar,Iassumedthatitwasprobablytrue.Ifitlookednew(orimprobably
extreme),Irejectedit.TheimpressionoffamiliarityisproducedbySystem1,andSystem
2reliesonthatimpressionforatrue/falsejudgment.
Thelessonoffigure5isthatpredictableillusionsinevitablyoccurifajudgmentis
basedonanimpressionofcognitiveeaseorstrain.Anythingthatmakesiteasierforthe
associativemachinetorunsmoothlywillalsobiasbeliefs.Areliablewaytomakepeople
believeinfalsehoodsisfrequentrepetition,becausefamiliarityisnoteasilydistinguished
fromtruth.Authoritarianinstitutionsand marketershavealwaysknown this fact.Butit
waspsychologistswhodiscoveredthatyoudonothavetorepeattheentirestatementofa
factorideatomakeitappeartrue.Peoplewhowererepeatedlyexposedtothephrase“the
bodytemperatureofachicken”weremorelikelytoacceptastruethestatementthat“the
bodytemperatureofachickenis144°”(oranyotherarbitrarynumber).Thefamiliarityof
one phrase in the statement sufficed to make the whole statement feel familiar, and
therefore true. If you cannot remember the source of a statement, and have no way to
relateittootherthingsyouknow,youhavenooptionbuttogowiththesenseofcognitive
ease.
HowtoWriteaPersuasiveMessage
Supposeyoumustwriteamessagethatyouwanttherecipientstobelieve.Ofcourse,your
messagewillbetrue,butthatisnotnecessarilyenoughforpeopletobelievethatitistrue.
Itisentirelylegitimateforyoutoenlistcognitiveeasetoworkinyourfavor,andstudies
oftruthillusionsprovidespecificsuggestionsthatmayhelpyouachievethisgoal.
Thegeneralprincipleisthatanythingyoucandotoreducecognitivestrainwillhelp,
soyoushouldfirstmaximizelegibility.Comparethesetwostatements:
AdolfHitlerwasbornin1892.
AdolfHitlerwasbornin1887.
Botharefalse(Hitlerwasbornin1889),butexperimentshaveshownthatthefirstismore
likelytobebelieved.Moreadvice:ifyourmessageistobeprinted,usehigh-qualitypaper
tomaximizethecontrastbetweencharactersandtheirbackground.Ifyouusecolor,you
aremorelikelytobebelievedifyourtextisprintedinbrightblueorredthaninmiddling
shadesofgreen,yellow,orpaleblue.
Ifyoucareaboutbeingthoughtcredibleandintelligent,donotusecomplexlanguage
wheresimplerlanguagewilldo.MyPrincetontoncolleagueDannyOppenheimerrefuted
amythprevalentawotoncolmongundergraduatesaboutthevocabularythatprofessors
find most impressive. In an article titled “Consequences of Erudite Vernacular Utilized
IrrespectiveofNecessity:ProblemswithUsingLongWordsNeedlessly,”heshowedthat
couchingfamiliarideasinpretentiouslanguageistakenasasignofpoorintelligenceand
lowcredibility.
Inadditiontomakingyourmessagesimple,trytomakeitmemorable.Putyourideas
inverseifyoucan;theywillbemorelikelytobetakenastruth.Participantsinamuch
citedexperimentreaddozensofunfamiliaraphorisms,suchas:
Woesunitefoes.
Littlestrokeswilltumblegreatoaks.
Afaultconfessedishalfredressed.
Otherstudentsreadsomeofthesameproverbstransformedintononrhymingversions:
Woesuniteenemies.
Littlestrokeswilltumblegreattrees.
Afaultadmittedishalfredressed.
Theaphorismswerejudgedmoreinsightfulwhentheyrhymedthanwhentheydidnot.
Finally, if you quote a source, choose one with a name that is easy to pronounce.
Participants in an experiment were asked to evaluate the prospects of fictitious Turkish
companiesonthebasisofreportsfromtwobrokeragefirms.Foreachstock,oneofthe
reportscamefromaneasilypronouncedname(e.g.,Artan)andtheotherreportcamefrom
a firm with an unfortunate name (e.g., Taahhut). The reports sometimes disagreed. The
bestprocedurefortheobserverswouldhavebeentoaveragethetworeports,butthisis
not what they did. They gave much more weight to the report from Artan than to the
reportfromTaahhut.RememberthatSystem2islazyandthatmentaleffortisaversive.If
possible, the recipients of your message want to stay away from anything that reminds
themofeffort,includingasourcewithacomplicatedname.
Allthisisverygoodadvice,butweshouldnotgetcarriedaway.High-qualitypaper,
brightcolors,andrhymingorsimplelanguagewillnotbemuchhelpifyourmessageis
obviouslynonsensical,orifitcontradictsfactsthatyouraudienceknowstobetrue.The
psychologistswhodotheseexperimentsdonotbelievethatpeoplearestupidorinfinitely
gullible.Whatpsychologistsdobelieveisthatallofuslivemuchofourlifeguidedbythe
impressions of System 1—and we often do not know the source of these impressions.
Howdoyouknowthatastatementistrue?Ifitisstronglylinkedbylogicorassociationto
otherbeliefsorpreferencesyouhold,orcomesfromasourceyoutrustandlike,youwill
feel a sense of cognitive ease. The trouble is that there may be other causes for your
feelingofease—includingthequalityofthefontandtheappealingrhythmoftheprose—
andyouhavenosimplewayoftracingyourfeelingstotheirsource.Thisisthemessage
offigure5:thesenseofeaseorstrainhasmultiplecauses,anditisdifficulttoteasethem
apart.Difficult,butnotimpossible.Peoplecanovercomesomeofthesuperficialfactors
that produce illusions of truth when strongly motivated to do so. On most occasions,
however,thelazySystem2willadoptthesuggestionsofSystem1andmarchon.
StrainandEffort
Thesymmetryofmanyassociativeconnectionswasadominantthemeinthediscussionof
associativecoherence.Aswesawearlier,peoplewhoaremadeto“smile”or“frown”by
stickingapencilintheirmouthorholdingaballbetweentheirfurrowedbrowsareprone
toexperience the emotionsthat frowningand smilingnormallyexpress. Thesame self-
reinforcingreciprocityisfoundinstudiesofcognitiveease.Ontheonehand,cognitive
strainisexperiencedwhentheeffortfuloperationsofSystem2areengaged.Ontheother
hand,theexperienceofcognitivestrain,whateveritssource,tendstomobilizeSystem2,
shiftingpeople’s approach to problemsfromacasualintuitivemodetoamoreengaged
andanalyticmode.
The bat-and-ball problem was mentioned earlier as a test of people’s tendency to
answerquestionswiththefirstideathatcomestotheirmind,withoutcheckingit.Shane
Frederick’sCognitiveReflectionTestconsistsofthebat-and-ballproblemandtwoothers,
allchosenbecausetheyevokeanimmediateintuitiveanswerthatisincorrect.Theother
twoitemsintheCRTare:
If it takes 5 machines 5 minutes to make 5 widgets, how long would it take 100
machinestomake100widgets?
100minutesOR5minutes
Inalake,thereisapatchoflilypads.Everyday,thepatchdoublesinsize.
Ifittakes48daysforthepatchtocovertheentirelake,howlongwouldittakefor
thepatchtocoverhalfofthelake?
24daysOR47days
Thecorrectanswersto both problemsareina footnote at thebottomofthe page.*The
experimenters recruited 40 Princeton students to take the CRT. Half of them saw the
puzzlesinasmallfontinwashed-outgrayprint.Thepuzzleswerelegible,butthe font
inducedcognitivestrain.Theresultstellaclearstory:90%ofthestudentswhosawthe
CRTinnormalfontmadeatleastonemistakeinthetest,buttheproportiondroppedto
35% when the font was barely legible. You read this correctly: performance was better
with the bad font. Cognitive strain, whatever its source, mobilizes System 2, which is
morelikelytorejecttheintuitiveanswersuggestedbySystem1.
ThePleasureofCognitiveEase
An article titled “Mind at Ease Puts a Smile on the Face” describes an experiment in
which participants were briefly shown pictures of objects. Some of these pictures were
made easier to recognize by showing the outline of the object just before the complete
image was shown, so briefly that the contours were never noticed. Emotional reactions
weremeasuredbyrecordingelectricalimpulsesfromfacialmuscles,registeringchanges
ofexpressionthataretooslightandtoobrieftobedetectablebyobservers.Asexpected,
peopleshowed a faintsmileand relaxedbrows when thepictures were easierto see.It
appearstobeafeatureofSystem1thatcognitiveeaseisassociatedwithgoodfeelings.
As expected, easily pronounced words evoke a favorable attitude. Companies with
pronounceablenamesdmisorrectlobetterthanothersforthefirstweekafterthestockis
issued,thoughtheeffectdisappearsovertime.Stockswithpronounceabletradingsymbols
(likeKARorLUNMOO)outperformthosewithtongue-twistingtickerslikePXGorRDO
—and they appear to retain a small advantage over some time. A study conducted in
Switzerland found that investors believe that stocks with fluent names like Emmi,
Swissfirst,andCometwillearnhigherreturnsthanthosewithclunkylabelslikeGeberit
andYpsomed.
Aswesawinfigure5,repetitioninducescognitiveeaseandacomfortingfeelingof
familiarity. The famed psychologist Robert Zajonc dedicated much of his career to the
studyofthelinkbetweentherepetitionofanarbitrarystimulusandthemildaffectionthat
peopleeventuallyhaveforit.Zajonccalleditthemereexposureeffect.Ademonstration
conductedinthestudentnewspapersoftheUniversityofMichiganandofMichiganState
Universityisoneofmyfavoriteexperiments.Foraperiodofsomeweeks,anad-likebox
appearedonthefrontpageofthepaper,whichcontainedoneofthefollowingTurkish(or
Turkish-sounding)words:kadirga,saricik,biwonjni,nansoma,andiktitaf.Thefrequency
withwhichthewordswererepeatedvaried:oneofthewordswasshownonlyonce,the
othersappearedontwo,five,ten,ortwenty-fiveseparateoccasions.(Thewordsthatwere
presentedmostofteninoneoftheuniversitypapersweretheleastfrequentintheother.)
Noexplanationwasoffered,andreaders’querieswereansweredbythestatementthat“the
purchaserofthedisplaywishedforanonymity.”
Whenthemysteriousseriesofadsended,theinvestigatorssentquestionnairestothe
university communities, asking for impressions of whether each of the words “means
something‘good’orsomething‘bad.’”Theresultswerespectacular:thewordsthatwere
presentedmorefrequentlywereratedmuchmorefavorablythanthewordsthathadbeen
shownonlyonceortwice.The finding has been confirmedinmanyexperiments,using
Chineseideographs,faces,andrandomlyshapedpolygons.
Themereexposureeffectdoesnotdependontheconsciousexperienceoffamiliarity.
In fact, the effect does not depend on consciousness at all: it occurs even when the
repeatedwordsorpicturesareshownsoquicklythattheobserversneverbecomeawareof
havingseenthem.Theystillenduplikingthewordsorpicturesthatwerepresentedmore
frequently.Asshouldbeclearbynow,System1canrespondtoimpressionsofeventsof
which System 2 is unaware. Indeed, the mere exposure effect is actually stronger for
stimulithattheindividualneverconsciouslysees.
Zajonc argued that the effect of repetition on liking is a profoundly important
biological fact, and that it extends to all animals. To survive in a frequently dangerous
world,anorganismshouldreactcautiouslytoanovelstimulus,withwithdrawalandfear.
Survivalprospectsarepoorforananimalthatisnotsuspiciousofnovelty.However,itis
also adaptive for the initial caution to fade if the stimulus is actually safe. The mere
exposure effect occurs, Zajonc claimed, because the repeated exposure of a stimulus is
followed by nothing bad. Such a stimulus will eventually become a safety signal, and
safetyisgood.Obviously,thisargumentisnotrestrictedtohumans.Tomakethatpoint,
oneofZajonc’sassociatesexposedtwosetsoffertilechickeneggstodifferenttones.After
theyhatched,thechicksconsistentlyemittedfewerdistresscallswhenexposedtothetone
theyhadheardwhileinhabitingtheshell.
Zajoncofferedaneloquentsummaryofhingictsprogramofresearch:
Theconsequencesofrepeatedexposuresbenefittheorganisminitsrelationstothe
immediate animate and inanimate environment. They allow the organism to
distinguishobjectsandhabitatsthataresafefromthosethatarenot,andtheyarethe
mostprimitivebasisofsocialattachments.Therefore,theyformthebasisforsocial
organizationandcohesion—thebasicsourcesofpsychologicalandsocialstability.
ThelinkbetweenpositiveemotionandcognitiveeaseinSystem1hasalongevolutionary
history.
Ease,Mood,andIntuition
Around1960,ayoungpsychologistnamedSarnoffMednickthoughthehadidentifiedthe
essenceofcreativity.Hisideawasassimpleasitwaspowerful:creativityisassociative
memorythatworksexceptionallywell.Hemadeupatest,calledtheRemoteAssociation
Test(RAT),whichisstilloftenusedinstudiesofcreativity.
Foraneasyexample,considerthefollowingthreewords:
cottageSwisscake
Canyouthinkofawordthatisassociatedwithallthree?Youprobablyworkedoutthat
theanswerischeese.Nowtrythis:
divelightrocket
Thisproblemismuchharder,butithasauniquecorrectanswer,whicheveryspeakerof
English recognizes, although less than 20% of a sample of students found it within 15
seconds. The answer is sky. Of course, not every triad of words has a solution. For
example,thewordsdream,ball,bookdonothaveasharedassociationthateveryonewill
recognizeasvalid.
Several teams of German psychologists that have studied the RAT in recent years
havecomeupwithremarkablediscoveriesaboutcognitiveease.Oneoftheteamsraised
twoquestions:Canpeoplefeelthatatriadofwordshasasolutionbeforetheyknowwhat
thesolutionis?Howdoesmoodinfluenceperformanceinthistask?Tofindout,theyfirst
made some of their subjects happy and others sad, by asking them to think for several
minutesabouthappyorsadepisodesintheirlives.Thentheypresentedthesesubjectswith
aseriesoftriads,halfofthemlinked(suchasdive,light,rocket)andhalfunlinked(such
as dream, ball, book), and instructed them to press one of two keys very quickly to
indicatetheirguessaboutwhetherthetriadwaslinked.Thetimeallowedforthisguess,2
seconds,wasmuchtooshortfortheactualsolutiontocometoanyone’smind.
Thefirstsurpriseisthatpeople’sguessesaremuchmoreaccuratethantheywouldbe
bychance.Ifindthisastonishing.Asenseofcognitiveeaseisapparentlygeneratedbya
veryfaintsignalfromthe associative machine, which“knows”thatthethreewords are
coherent (share an association) long before the association is retrieved. The role of
cognitiveeaseinthejudgmentwasconfirmedexperimentallybyanotherGermanteam:
manipulationsthatincreasecognitiveease(priming,aclearfont,pre-exposingwords)all
increasethetendencytoseethewordsaslinked.
Another remarkable discovery is the powerful effect of mood on this intuitive
performance. The experimentershape tende computed an “intuition index” to measure
accuracy. They found that putting the participants in a good mood before the test by
having them think happy thoughts more than doubled accuracy. An even more striking
resultisthatunhappysubjectswerecompletelyincapableofperformingtheintuitivetask
accurately;theirguesseswerenobetterthanrandom.Moodevidentlyaffectstheoperation
ofSystem1:whenweareuncomfortableandunhappy,welosetouchwithourintuition.
These findings add to the growing evidence that good mood, intuition, creativity,
gullibility,andincreasedrelianceonSystem1formacluster.Attheotherpole,sadness,
vigilance,suspicion,ananalyticapproach,andincreasedeffortalsogotogether.Ahappy
moodloosensthecontrolofSystem2overperformance:wheninagoodmood,people
becomemoreintuitiveandmorecreativebutalsolessvigilantandmorepronetological
errors.Hereagain,asinthemereexposureeffect,theconnectionmakesbiologicalsense.
Agoodmoodisasignalthatthingsaregenerallygoingwell,theenvironmentissafe,and
itisallrighttoletone’sguarddown.Abadmoodindicatesthatthingsarenotgoingvery
well,theremaybeathreat,andvigilanceisrequired.Cognitiveeaseisbothacauseanda
consequenceofapleasantfeeling.
The Remote Association Test has more to tell us about the link between cognitive
easeandpositiveaffect.Brieflyconsidertwotriadsofwords:
sleepmailswitch
saltdeepfoam
Youcouldnotknowit,ofcourse,butmeasurementsofelectricalactivityinthemusclesof
your face would probably have shown a slight smile when you read the second triad,
which is coherent (sea is the solution). This smiling reaction to coherence appears in
subjects who are told nothing about common associates; they are merely shown a
verticallyarrangedtriadofwordsandinstructedtopressthespacebaraftertheyhaveread
it.Theimpressionofcognitiveeasethatcomeswiththepresentationofacoherenttriad
appearstobemildlypleasurableinitself.
Theevidencethatwehaveaboutgoodfeelings,cognitiveease,andtheintuitionof
coherenceis,asscientistssay,correlationalbutnotnecessarilycausal.Cognitiveeaseand
smilingoccurtogether,butdothegoodfeelingsactuallyleadtointuitionsofcoherence?
Yes, they do. The proof comes from a clever experimental approach that has become
increasingly popular. Some participants were given a cover story that provided an
alternativeinterpretationfortheirgoodfeeling:theyweretoldaboutmusicplayedintheir
earphones that “previous research showed that this music influences the emotional
reactionsofindividuals.”Thisstorycompletelyeliminatestheintuitionofcoherence.The
findingshowsthatthebriefemotionalresponsethatfollowsthepresentationofatriadof
words (pleasant if the triad is coherent, unpleasant otherwise) is actually the basis of
judgments of coherence. There is nothing here that System 1 cannot do. Emotional
changesarenowexpected,andbecausetheyareunsurprisingtheyarenotlinkedcausally
tothewords.
This is as good as psychological research ever gets, in its combination of
experimentaltechniquesandinitsresults,whicharebothrobustandextremelysurprising.
We have learned a great deal about the automatic workings of System 1 in the last
decades.Muchofwhatwenowknowwouldhavesoundedlikesciencefictionthirtyor
fortyyearsago.Itwasbeyondimaginingthatbadfontinfluencesjudgmentsoftruthand
improvescognitiveperformance,orthatanemotionalresponsetothecognitiveeaseofa
triprthataadofwordsmediatesimpressionsofcoherence.Psychologyhascomealong
way.
SpeakingofCognitiveEase
“Let’snotdismisstheirbusinessplanjustbecausethefontmakesithardtoread.”
“We must be inclined to believe it because it has been repeated so often, but let’s
thinkitthroughagain.”
“Familiaritybreedsliking.Thisisamereexposureeffect.”
“I’minaverygoodmoodtoday,andmySystem2isweakerthanusual.Ishouldbe
extracareful.”
P
Norms,Surprises,andCauses
The central characteristics and functions of System 1 and System 2 have now been
introduced, with a more detailed treatment of System 1. Freely mixing metaphors, we
have in our head a remarkably powerful computer, not fast by conventional hardware
standards,butabletorepresentthestructureofourworldbyvarioustypesofassociative
links in a vast network of various types of ideas. The spreading of activation in the
associative machine is automatic, but we (System 2) have some ability to control the
search of memory, and also to program it so that the detection of an event in the
environment can attract attention. We next go into more detail of the wonders and
limitationofwhatSystem1cando.
AssessingNormality
ThemainfunctionofSystem1istomaintainandupdateamodelofyourpersonalworld,
whichrepresentswhatisnormalinit.Themodelisconstructedbyassociationsthatlink
ideasofcircumstances,events,actions,andoutcomesthatco-occurwithsomeregularity,
eitheratthesametimeorwithinarelativelyshortinterval.Astheselinksareformedand
strengthened,thepatternofassociatedideascomestorepresentthestructureofeventsin
yourlife,anditdeterminesyourinterpretationofthepresentaswellasyourexpectations
ofthefuture.
Acapacityforsurpriseisanessentialaspectofourmentallife,andsurpriseitselfis
themostsensitiveindicationofhowweunderstandourworldandwhatweexpectfromit.
Therearetwomainvarietiesofsurprise.Someexpectationsareactiveandconscious—
youknowyouarewaitingforaparticularevent to happen. When the hour is near,you
maybeexpectingthesoundofthedoorasyourchildreturnsfromschool;whenthedoor
opens you expect the sound of a familiar voice. You will be surprised if an actively
expected event does not occur. But there is a much larger category of events that you
expectpassively;youdon’twaitforthem,butyouarenotsurprisedwhentheyhappen.
These are events that are normal in a situation, though not sufficiently probable to be
activelyexpected.
Asingleincidentmaymakea recurrencelesssurprising.Someyearsago, my wife
and I were of dealWhen normvacationing in a small island resort on the Great Barrier
Reef.Thereareonlyfortyguestroomsontheisland.Whenwecametodinner,wewere
surprised to meet an acquaintance, a psychologist named Jon. We greeted each other
warmlyandcommentedonthecoincidence.Jonlefttheresort the next day.Abouttwo
weekslater,wewereinatheaterinLondon.Alatecomersatnexttomeafterthelights
wentdown.Whenthelightscameupfortheintermission,IsawthatmyneighborwasJon.
MywifeandIcommentedlaterthatweweresimultaneouslyconsciousoftwofacts:first,
thiswasamoreremarkablecoincidencethanthefirstmeeting;second,weweredistinctly
lesssurprisedtomeetJononthesecondoccasionthanwehadbeenonthefirst.Evidently,
thefirstmeetinghadsomehowchangedtheideaofJoninourminds.Hewasnow“the
psychologist who shows up when we travel abroad.” We (System 2) knew this was a
ludicrousidea,butourSystem1hadmadeitseemalmostnormaltomeetJoninstrange
places.Wewouldhaveexperiencedmuchmoresurpriseifwehadmetanyacquaintance
otherthanJoninthenextseatofaLondontheater.Byanymeasureofprobability,meeting
Jon in the theater was much less likely than meeting any one of our hundreds of
acquaintances—yetmeetingJonseemedmorenormal.
Under some conditions, passive expectations quickly turn active, as we found in
anothercoincidence.On a Sunday evening someyearsago,weweredrivingfromNew
York City to Princeton, as we had been doing every week for a longtime. We saw an
unusualsight:acaronfirebythesideoftheroad.Whenwereachedthesamestretchof
roadthefollowingSunday,anothercarwasburningthere.Hereagain,wefoundthatwe
weredistinctlylesssurprisedonthesecondoccasionthanwehadbeenonthefirst.This
wasnow“theplacewherecarscatchfire.”Becausethecircumstancesoftherecurrence
were the same, the second incident was sufficient to create an active expectation: for
months,perhapsforyears,aftertheeventwewereremindedofburningcarswheneverwe
reachedthatspotoftheroadandwerequitepreparedtoseeanotherone(butofcoursewe
neverdid).
ThepsychologistDaleMillerandIwroteanessayinwhichweattemptedtoexplain
howeventscometobeperceivedasnormalorabnormal.Iwilluseanexamplefromour
descriptionof“normtheory,”althoughmyinterpretationofithaschangedslightly:
An observer, casually watching the patrons at a neighboring table in a fashionable
restaurant, notices that the first guest to taste the soup winces, as if in pain. The
normality of a multitude of events will be altered by this incident. It is now
unsurprisingfortheguestwhofirsttastedthesouptostartleviolentlywhentouched
byawaiter;itisalsounsurprisingforanotherguesttostifleacrywhentastingsoup
fromthesametureen.Theseeventsandmanyothersappearmorenormalthanthey
would have otherwise, but not necessarily because they confirm advance
expectations. Rather, they appear normal because they recruit the original episode,
retrieveitfrommemory,andareinterpretedinconjunctionwithit.
Imagineyourselftheobserverattherestaurant.Youweresurprisedbythefirstguest’s
unusualreactiontothesoup,andsurprisedagainbythestartledresponsetothewaiter’s
touch.However,thesecondabnormaleventwillretrievethefirstfrommemory,andboth
make sense together. The two events fit into a pattern, in which the guest is an
exceptionallytenseperson.Ontheotherhand,ifthenextthingthathappensafterthefirst
guest’s grimace is that another customer rejects the soup, these two surprises will be
linkedandthehinsursoupwillsurelybeblamed.
“HowmanyanimalsofeachkinddidMosestakeintotheark?”Thenumberofpeople
whodetectwhatiswrongwiththisquestionissosmallthatithasbeendubbedthe“Moses
illusion.”Mosestooknoanimalsintotheark;Noahdid.Liketheincidentofthewincing
soupeater,theMosesillusionisreadilyexplainedbynormtheory.Theideaofanimals
goingintothearksetsupabiblicalcontext,andMosesisnotabnormalinthatcontext.
Youdidnotpositivelyexpecthim,butthementionofhisnameisnotsurprising.Italso
helpsthatMosesandNoahhavethesamevowelsoundandnumberofsyllables.Aswith
the triads that produce cognitive ease, you unconsciously detect associative coherence
between “Moses” and “ark” and so quickly accept the question. Replace Moses with
GeorgeW.Bushinthissentenceandyouwillhaveapoorpoliticaljokebutnoillusion.
Whensomethingcementdoesnotfitintothecurrentcontextofactivatedideas,the
systemdetectsanabnormality,asyoujustexperienced.Youhadnoparticularideaofwhat
was coming after something, but you knew when the word cement came that it was
abnormal in that sentence. Studies of brain responses have shown that violations of
normalityaredetectedwithastonishingspeedandsubtlety.Inarecentexperiment,people
heardthesentence“Earth revolves aroundthetrouble every year.”Adistinctivepattern
wasdetectedinbrainactivity,startingwithintwo-tenthsofasecondoftheonsetofthe
oddword.Evenmoreremarkable,thesamebrainresponseoccursatthesamespeedwhen
amalevoicesays,“IbelieveIampregnantbecauseIfeelsickeverymorning,”orwhenan
upper-class voice says, “I have a large tattoo on my back.” A vast amount of world
knowledge must instantly be brought to bear for the incongruity to be recognized: the
voicemustbeidentifiedasupper-classEnglishandconfrontedwiththegeneralizationthat
largetattoosareuncommonintheupperclass.
Weareabletocommunicatewitheachotherbecauseourknowledgeoftheworldand
ouruseofwordsarelargelyshared.WhenImentionatable,withoutspecifyingfurther,
you understand that I mean a normal table. You know with certainty that its surface is
approximately level and that it has far fewer than 25 legs. We have norms for a vast
number of categories, and these norms provide the background for the immediate
detectionofanomaliessuchaspregnantmenandtattooedaristocrats.
Toappreciatetheroleofnormsincommunication,considerthesentence“Thelarge
mouse climbed over the trunk of the very small elephant.” I can count on your having
normsforthesizeofmiceandelephantsthatarenottoofarfrommine.Thenormsspecify
atypicalor average sizeforthese animals, andtheyalso contain informationaboutthe
rangeorvariabilitywithinthecategory.Itisveryunlikelythateitherofusgottheimage
inourmind’seyeofamouselargerthananelephantstridingoveranelephantsmallerthan
amouse.Instead,weeachseparatelybutjointlyvisualizedamousesmallerthanashoe
clamberingoveranelephantlargerthanasofa.System1,whichunderstandslanguage,has
accesstonormsofcategories,whichspecifytherangeofplausiblevaluesaswellasthe
mosttypicalcases.
SeeingCausesandIntentions
“Fred’sparentsarrivedlate.Thecatererswereexpectedsoon.Fredwasangry.”Youknow
why Fred was angry, and it is not because the caterers were expected soon. In your
networkofassociationsmalsinco,angerandlackofpunctualityarelinked asaneffect
anditspossiblecause,butthereisnosuchlinkbetweenangerandtheideaofexpecting
caterers.Acoherentstorywasinstantlyconstructedasyouread;youimmediatelyknew
thecauseofFred’sanger.Findingsuchcausalconnectionsispartofunderstandingastory
andisanautomaticoperationofSystem1.System2,yourconsciousself,wasofferedthe
causalinterpretationandacceptedit.
A story in Nassim Taleb’s The Black Swan illustrates this automatic search for
causality. He reports that bond prices initially rose on the day of Saddam Hussein’s
capture in his hiding place in Iraq. Investors were apparently seeking safer assets that
morning, and the Bloomberg News service flashed this headline: U.S. TREASURIES RISE;
HUSSEINCAPTUREMAYNOTCURBTERRORISM.Halfanhourlater,bondpricesfellbackand
therevisedheadlineread:U.S.TREASURIESFALL;HUSSEINCAPTUREBOOSTSALLUREOFRISKY
ASSETS.Obviously,Hussein’scapturewasthemajoreventoftheday,andbecauseofthe
waytheautomaticsearchforcausesshapesourthinking,thateventwasdestinedtobethe
explanation of whatever happened in the market on that day. The two headlines look
superficiallylikeexplanationsofwhathappenedinthemarket,butastatementthatcan
explaintwocontradictoryoutcomesexplainsnothingatall.Infact,alltheheadlinesdois
satisfy our need for coherence: a large event is supposed to have consequences, and
consequences need causes to explain them. We have limited information about what
happenedonaday,andSystem1isadeptatfindingacoherentcausalstorythatlinksthe
fragmentsofknowledgeatitsdisposal.
Readthissentence:
AfterspendingadayexploringbeautifulsightsinthecrowdedstreetsofNewYork,
Janediscoveredthatherwalletwasmissing.
Whenpeoplewhohadreadthisbriefstory(alongwithmanyothers)weregivenasurprise
recalltest,thewordpickpocketwasmorestronglyassociatedwiththestorythantheword
sights,eventhoughthelatterwasactuallyinthesentencewhiletheformerwasnot.The
rules of associative coherence tell us what happened. The event of a lost wallet could
evokemanydifferentcauses:thewalletslippedoutofapocket,wasleftintherestaurant,
etc.However,whentheideasoflostwallet,NewYork,andcrowdsarejuxtaposed,they
jointlyevoketheexplanationthatapickpocketcausedtheloss.Inthestoryofthestartling
soup,theoutcome—whetheranothercustomerwincingatthetasteofthesouporthefirst
person’s extreme reaction to the waiter’s touch—brings about an associatively coherent
interpretationoftheinitialsurprise,completingaplausiblestory.
The aristocratic Belgian psychologist Albert Michotte published a book in 1945
(translated into English in 1963) that overturned centuries of thinking about causality,
going back at least to Hume’s examination of the association of ideas. The commonly
accepted wisdom was that we infer physical causality from repeated observations of
correlationsamongevents.Wehavehadmyriadexperiencesinwhichwesawoneobject
in motion touching another object, which immediately starts to move, often (but not
always)inthesamedirection.Thisiswhathappenswhenabilliardballhitsanother,andit
isalsowhathappenswhenyouknockoveravasebybrushingagainstit.Michottehada
differentidea:hearguedthatweseecausality,justasdirectlyasweseecolor.Tomakehis
point,hecreatedepisodesinnttiowhichablacksquaredrawnonpaperisseeninmotion;
it comes into contact with another square, which immediately begins to move. The
observers know that there is no real physical contact, but they nevertheless have a
powerful“illusionofcausality.”Ifthesecondobjectstartsmovinginstantly,theydescribe
it as having been “launched” by the first. Experiments have shown that six-month-old
infantssee thesequence ofevents as a cause-effectscenario, and they indicate surprise
whenthesequenceisaltered. Weareevidentlyreadyfrombirthtohaveimpressionsof
causality, which do not depend on reasoning about patterns of causation. They are
productsofSystem1.
In1944,ataboutthesametimeasMichottepublishedhisdemonstrationsofphysical
causality,thepsychologistsFritzHeiderandMary-AnnSimmelusedamethodsimilarto
Michotte’stodemonstratetheperceptionofintentionalcausality.Theymadeafilm,which
lasts all of one minute and forty seconds, in which you see a large triangle, a small
triangle,andacirclemovingaroundashapethatlookslikeaschematicviewofahouse
withanopendoor.Viewersseeanaggressivelargetrianglebullyingasmallertriangle,a
terrifiedcircle,thecircleandthesmalltrianglejoiningforcestodefeatthebully;theyalso
observemuchinteractionaroundadoorandthenanexplosivefinale.Theperceptionof
intentionandemotionisirresistible;onlypeopleafflictedbyautismdonotexperienceit.
Allthisisentirelyinyourmind,ofcourse.Yourmindisreadyandeveneagertoidentify
agents, assign them personality traits and specific intentions, and view their actions as
expressingindividualpropensities.Hereagain,theevidenceisthatwearebornprepared
tomakeintentionalattributions:infants under oneyearold identify bulliesandvictims,
andexpectapursuertofollowthemostdirectpathinattemptingtocatchwhateveritis
chasing.
The experience of freely willed action is quite separate from physical causality.
Althoughitisyourhandthatpicksupthesalt,youdonotthinkoftheeventintermsofa
chainofphysicalcausation.Youexperienceitascausedbyadecisionthatadisembodied
youmade,becauseyouwanted to add salttoyourfood.Many people finditnaturalto
describe their soul as the source and the cause of their actions. The psychologist Paul
Bloom,writinginTheAtlanticin2005,presentedtheprovocativeclaimthatourinborn
readinessto separatephysical andintentional causalityexplains thenear universalityof
religious beliefs. He observes that “we perceive the world of objects as essentially
separatefromtheworldofminds,makingitpossibleforustoenvisionsoullessbodiesand
bodilesssouls.”Thetwomodesofcausationthatwearesettoperceivemakeitnaturalfor
us to accept the two central beliefs of many religions: an immaterial divinity is the
ultimatecauseofthephysicalworld,andimmortalsoulstemporarilycontrolourbodies
while we live and leave them behind as we die. In Bloom’s view, the two concepts of
causalitywereshapedseparatelybyevolutionaryforces,buildingtheoriginsofreligion
intothestructureofSystem1.
Theprominenceofcausalintuitionsisarecurrentthemeinthisbookbecausepeople
are prone to apply causal thinking inappropriately, to situations that require statistical
reasoning.Statisticalthinkingderivesconclusionsaboutindividualcasesfromproperties
ofcategoriesandensembles.Unfortunately,System1doesnothavethecapabilityforthis
modeofreasoning;System2canlearntothink statistically,butfewpeoplereceivethe
necessarytraining.
The psychology of causality was the basis of my decision to describe psycl c to
thinhological processes by metaphors of agency, with little concern for consistency. I
sometimesrefertoSystem1asanagentwithcertaintraitsandpreferences,andsometimes
asanassociativemachinethatrepresentsrealitybyacomplexpatternoflinks.Thesystem
andthemachinearefictions;myreasonforusingthemisthattheyfitthewaywethink
aboutcauses.Heider’strianglesandcirclesarenotreallyagents—itisjustveryeasyand
naturaltothinkofthemthatway.Itisamatterofmentaleconomy.Iassumethatyou(like
me)finditeasiertothinkaboutthemindifwedescribewhathappensintermsoftraits
andintentions(thetwosystems)andsometimesintermsofmechanicalregularities(the
associativemachine).Idonotintendtoconvinceyouthatthesystemsarereal,anymore
thanHeiderintendedyoutobelievethatthelargetriangleisreallyabully.
SpeakingofNormsandCauses
“Whenthesecondapplicantalsoturnedouttobeanoldfriendofmine,Iwasn’tquite
assurprised.Verylittlerepetitionisneededforanewexperiencetofeelnormal!”
“When we survey the reaction to these products, let’s make sure we don’t focus
exclusivelyontheaverage.Weshouldconsidertheentirerangeofnormalreactions.”
“Shecan’tacceptthatshewasjustunlucky;sheneedsacausalstory.Shewillendup
thinkingthatsomeoneintentionallysabotagedherwork.”
P
AMachineforJumpingtoConclusions
ThegreatcomedianDannyKayehadalinethathasstayedwithmesincemyadolescence.
Speakingofawomanhedislikes,hesays,“Herfavoritepositionisbesideherself,andher
favorite sport is jumping to conclusions.” The line came up, I remember, in the initial
conversationwithAmosTverskyabouttherationalityofstatisticalintuitions,andnowI
believeitoffersanaptdescriptionofhowSystem1functions.Jumpingtoconclusionsis
efficientiftheconclusionsarelikelytobecorrectandthecostsofanoccasionalmistake
acceptable,andifthejumpsavesmuchtimeandeffort.Jumpingtoconclusionsisrisky
whenthesituationisunfamiliar,thestakesarehigh,andthereisnotimetocollectmore
information. These are the circumstances in which intuitive errors are probable, which
maybepreventedbyadeliberateinterventionofSystem2.
NeglectofAmbiguityandSuppressionofDoubt
Figure6
What do the three exhibits in figure 6 have in common? The answer is that all are
ambiguous.YoualmostcertainlyreadthedisplayontheleftasABCandtheoneonthe
rightas121314,butthemiddleitemsinbothdisplaysareidentical.Youcouldjustas
wellhavereadeiomprthecvethemasA13Cor12B14,butyoudidnot.Whynot?The
same shape is read as a letter in a context of letters and as a number in a context of
numbers.Theentirecontexthelpsdeterminetheinterpretationofeachelement.Theshape
isambiguous,butyoujumptoaconclusionaboutitsidentityanddonotbecomeawareof
theambiguitythatwasresolved.
As for Ann, you probably imagined a woman with money on her mind, walking
towardabuildingwithtellersandsecurevaults.Butthisplausibleinterpretationisnotthe
onlypossibleone;thesentenceisambiguous.Ifanearliersentencehadbeen“Theywere
floatinggentlydowntheriver,”youwouldhaveimaginedanaltogetherdifferentscene.
Whenyouhavejustbeenthinkingofariver,thewordbankisnotassociatedwithmoney.
Intheabsenceofanexplicitcontext,System1generatedalikelycontextonitsown.We
knowthatitisSystem1becauseyouwerenotawareofthechoiceorofthepossibilityof
anotherinterpretation.Unlessyouhavebeencanoeingrecently,youprobablyspendmore
timegoingtobanksthanfloatingonrivers,andyouresolvedtheambiguityaccordingly.
Whenuncertain,System1betsonananswer,andthebetsareguidedbyexperience.The
rules of the betting are intelligent: recent events and the current context have the most
weight in determining an interpretation. When no recent event comes to mind, more
distant memories govern. Among your earliest and most memorable experiences was
singingyourABCs;youdidnotsingyourA13Cs.
Themostimportantaspectofbothexamplesisthatadefinitechoicewasmade,but
youdidnotknowit.Onlyoneinterpretationcametomind,andyouwereneverawareof
theambiguity.System1doesnotkeeptrackofalternativesthatitrejects,orevenofthe
factthattherewerealternatives.ConsciousdoubtisnotintherepertoireofSystem1;it
requires maintaining incompatible interpretations in mind at the same time, which
demandsmentaleffort.UncertaintyanddoubtarethedomainofSystem2.
ABiastoBelieveandConfirm
ThepsychologistDanielGilbert,widelyknownastheauthorofStumblingtoHappiness,
once wrote an essay, titled “How Mental Systems Believe,” in which he developed a
theoryofbelievingandunbelievingthathetracedtotheseventeenth-centuryphilosopher
Baruch Spinoza. Gilbert proposed that understanding a statement must begin with an
attempttobelieveit:youmustfirstknowwhattheideawouldmeanifitweretrue.Only
then can you decide whether or not to unbelieve it. The initial attempt to believe is an
automatic operation of System 1, which involves the construction of the best possible
interpretation of the situation. Even a nonsensical statement, Gilbert argues, will evoke
initialbelief.Tryhisexample:“whitefisheatcandy.”Youprobablywereawareofvague
impressionsoffishandcandyasanautomaticprocessofassociativememorysearchedfor
linksbetweenthetwoideasthatwouldmakesenseofthenonsense.
Gilbert sees unbelieving as an operation of System 2, and he reported an elegant
experiment to make his point. The participants saw nonsensical assertions, such as “a
dincaisaflame,”followedafterafewsecondsbyasingleword,“true”or“false.”They
were later tested for their memory of which sentences had been labeled “true.” In one
conditionoftheexperimentsubjectswere required toholddigitsinmemoryduring the
task.ThedisruptionofSystem2hadaselectiveeffect:itmadeitdifficultforpeopleto
“unbelieve” false sentences. In a later test of memory, the depleted par muumbling
toticipants ended up thinking that many of the false sentences were true. The moral is
significant:whenSystem2isotherwiseengaged,wewillbelievealmostanything.System
1isgullibleandbiasedtobelieve,System2isinchargeofdoubtingandunbelieving,but
System2issometimesbusy,andoftenlazy.Indeed,thereisevidencethatpeoplearemore
likelytobeinfluencedbyemptypersuasivemessages,suchascommercials,whentheyare
tiredanddepleted.
The operations of associative memory contribute to a general confirmation bias.
Whenasked,“IsSamfriendly?”differentinstancesofSam’sbehaviorwillcometomind
thanwouldifyouhadbeenasked“IsSamunfriendly?”Adeliberatesearchforconfirming
evidence, known as positive test strategy, is also how System 2 tests a hypothesis.
Contrarytotherulesofphilosophersofscience,whoadvisetestinghypothesesbytrying
to refute them, people (and scientists, quite often) seek data that are likely to be
compatiblewiththebeliefstheycurrentlyhold.TheconfirmatorybiasofSystem1favors
uncritical acceptance of suggestions and exaggeration of the likelihood of extreme and
improbableevents.IfyouareaskedabouttheprobabilityofatsunamihittingCalifornia
withinthenextthirtyyears,theimagesthatcometoyourmindarelikelytobeimagesof
tsunamis,inthemannerGilbertproposedfornonsensestatementssuchas“whitefisheat
candy.”Youwillbepronetooverestimatetheprobabilityofadisaster.
ExaggeratedEmotionalCoherence(HaloEffect)
Ifyoulikethepresident’spolitics,youprobablylikehisvoiceandhisappearanceaswell.
Thetendencytolike(ordislike)everythingaboutaperson—includingthingsyouhavenot
observed—is known as the halo effect. The term has been in use in psychology for a
century,butithasnotcomeintowideuseineverydaylanguage.Thisisapity,becausethe
haloeffectisagoodnameforacommonbiasthatplaysalargeroleinshapingourviewof
peopleandsituations.ItisoneofthewaystherepresentationoftheworldthatSystem1
generatesissimplerandmorecoherentthantherealthing.
YoumeetawomannamedJoanatapartyandfindherpersonableandeasytotalkto.
Nowhernamecomesupassomeonewhocouldbeaskedtocontributetoacharity.What
do you know about Joan’s generosity? The correct answer is that you know virtually
nothing,becausethereislittlereasontobelievethatpeoplewhoareagreeableinsocial
situations are also generous contributors to charities. But you like Joan and you will
retrieve the feeling of liking her when you think of her. You also like generosity and
generous people. By association, you are now predisposed to believe that Joan is
generous.Andnowthatyoubelievesheisgenerous,youprobablylikeJoanevenbetter
thanyoudidearlier,becauseyouhaveaddedgenerositytoherpleasantattributes.
RealevidenceofgenerosityismissinginthestoryofJoan,andthegapisfilledbya
guessthatfitsone’semotionalresponsetoher.Inothersituations,evidenceaccumulates
graduallyandtheinterpretationisshapedbytheemotionattachedtothefirstimpression.
Inanenduringclassicofpsychology,SolomonAschpresenteddescriptionsoftwopeople
andaskedforcommentsontheirpersonality.WhatdoyouthinkofAlanandBen?
Alan:intelligent—industrious—impulsive—critical—stubborn—envious
Ben:envious—The#82stubborn—critical—impulsive—industrious—intelligent
Ifyouarelikemostofus,youviewedAlanmuchmorefavorablythanBen.Theinitial
traitsinthelistchangetheverymeaningofthetraitsthatappearlater.Thestubbornnessof
anintelligentpersonisseenaslikelytobejustifiedandmayactuallyevokerespect,but
intelligence in an envious and stubborn person makes him more dangerous. The halo
effect is also an example of suppressed ambiguity: like the word bank, the adjective
stubbornisambiguousandwillbeinterpretedinawaythatmakesitcoherentwiththe
context.
There have been many variations on this research theme. Participants in one study
firstconsideredthefirstthreeadjectivesthatdescribeAlan;thentheyconsideredthelast
three,whichbelonged,theyweretold,toanotherperson.Whentheyhadimaginedthetwo
individuals,theparticipantswereaskedifitwasplausibleforallsixadjectivestodescribe
thesameperson,andmostofthemthoughtitwasimpossible!
Thesequenceinwhichweobservecharacteristicsofapersonisoftendeterminedby
chance.Sequencematters,however,becausethehaloeffectincreasestheweightoffirst
impressions,sometimestothepointthatsubsequentinformationismostlywasted.Early
in my career as a professor, I graded students’ essay exams in the conventional way. I
wouldpickuponetestbookletatatimeandreadallthatstudent’sessaysinimmediate
succession,gradingthemasIwent.Iwouldthencomputethetotalandgoontothenext
student. I eventually noticed that my evaluations of the essays in each booklet were
strikinglyhomogeneous.Ibegantosuspectthatmygradingexhibitedahaloeffect,and
that the first question I scored had a disproportionate effect on the overall grade. The
mechanismwassimple:ifIhadgivenahighscoretothefirstessay,Igavethestudentthe
benefit of the doubt whenever I encountered a vague or ambiguous statement later on.
Thisseemedreasonable.Surelyastudentwhohaddonesowellonthefirstessaywould
notmakeafoolishmistakeinthesecondone!Buttherewasaseriousproblemwithmy
wayofdoingthings.Ifastudenthadwrittentwoessays,onestrongandoneweak,Iwould
endup withdifferentfinal gradesdepending onwhich essayI readfirst. I hadtold the
studentsthatthetwoessayshadequalweight,butthatwasnottrue:thefirstonehada
muchgreaterimpactonthefinalgradethanthesecond.Thiswasunacceptable.
Iadoptedanewprocedure.Insteadof reading the booklets insequence,Ireadand
scoredallthestudents’answerstothefirstquestion,thenwentontothenextone.Imade
suretowriteallthescoresontheinsidebackpageofthebookletsothatIwouldnotbe
biased(evenunconsciously)whenIreadthesecondessay.Soonafterswitchingtothenew
method,Imadeadisconcertingobservation:myconfidenceinmygradingwasnowmuch
lowerthanithadbeen.ThereasonwasthatIfrequentlyexperiencedadiscomfortthatwas
newtome.WhenIwasdisappointedwithastudent’ssecondessayandwenttotheback
pageofthebooklettoenterapoorgrade,IoccasionallydiscoveredthatIhadgivenatop
grade to the same student’s first essay. I also noticed that I was tempted to reduce the
discrepancybychangingthegradethatIhadnotyetwrittendown,andfoundithardto
followthesimpleruleofneveryieldingtothattemptation.Mygradesfortheessaysofa
single student often varied over a considerable range. The lack of coherence left me
uncertainandfrustrated.
IwasnowlesshappywithandlessconfidentinmygradesthanIhadbeenearlier,but
Irecognizedthatthassconfthiswasagoodsign,anindicationthatthenewprocedurewas
superior. The consistency I had enjoyed earlier was spurious; it produced a feeling of
cognitiveease,andmySystem2washappytolazilyacceptthefinalgrade.Byallowing
myself to be strongly influenced by the first question in evaluating subsequent ones, I
spared myself the dissonance of finding the same student doing very well on some
questionsandbadlyonothers.TheuncomfortableinconsistencythatwasrevealedwhenI
switched to the new procedure was real: it reflected both the inadequacy of any single
questionasameasureofwhatthestudentknewandtheunreliabilityofmyowngrading.
The procedure I adopted to tame the halo effect conforms to a general principle:
decorrelateerror!Tounderstandhowthisprincipleworks,imaginethatalargenumberof
observers are shown glass jars containing pennies and are challenged to estimate the
number of pennies in each jar. As James Surowiecki explained in his best-selling The
WisdomofCrowds,thisisthekindoftaskinwhichindividualsdoverypoorly,butpools
of individual judgments do remarkably well. Some individuals greatly overestimate the
truenumber,othersunderestimateit,butwhenmanyjudgmentsareaveraged,theaverage
tendstobequiteaccurate.Themechanismisstraightforward:allindividualslookatthe
samejar,andalltheirjudgmentshaveacommonbasis.Ontheotherhand,theerrorsthat
individualsmakeareindependentoftheerrorsmadebyothers,and(intheabsenceofa
systematicbias)theytendtoaveragetozero.However,themagicoferrorreductionworks
well only when the observations are independent and their errors uncorrelated. If the
observers share a bias, the aggregation of judgments will not reduce it. Allowing the
observerstoinfluenceeachothereffectivelyreducesthesizeofthesample,andwithitthe
precisionofthegroupestimate.
Toderivethemostusefulinformationfrommultiplesourcesofevidence,youshould
always try to make these sources independent of each other. This rule is part of good
policeprocedure.Whentherearemultiplewitnessestoanevent,theyarenotallowedto
discuss it before giving their testimony. The goal is not only to prevent collusion by
hostile witnesses, it is also to prevent unbiased witnesses from influencing each other.
Witnesses who exchange their experiences will tend to make similar errors in their
testimony, reducing the total value of the information they provide. Eliminating
redundancyfromyoursourcesofinformationisalwaysagoodidea.
The principle of independent judgments (and decorrelated errors) has immediate
applicationsfortheconductofmeetings,anactivityinwhichexecutivesinorganizations
spend a great deal of their working days. A simple rule can help: before an issue is
discussed,allmembersofthecommitteeshouldbeaskedtowriteaverybriefsummaryof
theirposition.Thisproceduremakesgooduseofthevalueofthediversityofknowledge
andopinioninthegroup.Thestandardpracticeofopendiscussiongivestoomuchweight
totheopinionsofthosewhospeakearlyandassertively,causingotherstolineupbehind
them.
WhatYouSeeisAllThereis(Wysiati)
One of my favorite memories of the early years of working with Amos is a comedy
routineheenjoyedperforming.Inaperfectimpersonationofoneoftheprofessorswith
whom he had studied philosophy as an undergraduate, Amos would growl in Hebrew
markedby athick Germanaccent: “Youmust neverforgetthe Primat of the Is.”What
exactly his teacher had meant by that phrase never became clear to me (or to Amos, I
believe),butAmos’sjokesalwaysmaht=cipdeapoint.Hewasremindedoftheoldphrase
(andeventuallyIwastoo)wheneverweencounteredtheremarkableasymmetrybetween
thewaysourmindtreatsinformationthatiscurrentlyavailableandinformationwedonot
have.
An essential design feature of the associative machine is that it represents only
activatedideas.Informationthatisnotretrieved(evenunconsciously)frommemorymight
aswellnotexist.System1excelsatconstructingthebestpossiblestorythatincorporates
ideascurrentlyactivated,butitdoesnot(cannot)allowforinformationitdoesnothave.
The measure of success for System 1 is the coherence of the story it manages to
create. The amount and quality of the data on which the story is based are largely
irrelevant.Wheninformationisscarce,whichisacommonoccurrence,System1operates
asamachineforjumpingtoconclusions.Considerthefollowing:“WillMindikbeagood
leader?Sheisintelligentandstrong…”Ananswerquicklycametoyourmind,anditwas
yes.Youpickedthebestanswerbasedontheverylimitedinformationavailable,butyou
jumpedthegun.Whatifthenexttwoadjectiveswerecorruptandcruel?
TakenoteofwhatyoudidnotdoasyoubrieflythoughtofMindikasaleader.You
didnotstartbyasking,“WhatwouldIneedtoknowbeforeIformedanopinionaboutthe
quality of someone’s leadership?” System 1 got to work on its own from the first
adjective:intelligentisgood,intelligentandstrongisverygood.Thisisthebeststorythat
can be constructed from two adjectives, and System 1 delivered it with great cognitive
ease.Thestorywillberevisedifnewinformationcomesin(suchasMindikiscorrupt),
but there is no waiting and no subjective discomfort. And there also remains a bias
favoringthefirstimpression.
Thecombinationofacoherence-seekingSystem1withalazySystem2impliesthat
System 2 will endorse many intuitive beliefs, which closely reflect the impressions
generated by System 1. Of course, System 2 also is capable of a more systematic and
carefulapproachtoevidence,andoffollowingalistofboxesthatmustbecheckedbefore
makingadecision—thinkofbuyingahome,whenyoudeliberatelyseekinformationthat
you don’t have. However, System 1 is expected to influence even the more careful
decisions.Itsinputneverceases.
Jumping to conclusions on the basis of limited evidence is so important to an
understandingofintuitivethinking,andcomesupsoofteninthisbook,thatIwillusea
cumbersomeabbreviationforit:WYSIATI,whichstandsforwhatyouseeisallthereis.
System1is radically insensitivetoboththe qualityandthequantity of theinformation
thatgivesrisetoimpressionsandintuitions.
Amos, with two of his graduate students at Stanford, reported a study that bears
directly on WYSIATI, by observing the reaction of people who are given one-sided
evidence and know it. The participants were exposed to legal scenarios such as the
following:
On September 3, plaintiff David Thornton, a forty-three-year-old union field
representative,waspresentinThriftyDrugStore#168,performingaroutineunion
visit.Withintenminutesofhisarrival,astoremanagerconfrontedhimandtoldhim
hecouldnolongerspeakwiththeunionemployeesonthefloorofthestore.Instead,
hewouldhavetoseetheminabackroomwhiletheywereonbreak.Sucharequest
is allowed by the union contract with Thrifty Drug but had never before been
enforced.WhenMr.Thorntonobjected,hewastoldthathehadthechoiceofconto
roomwhilformingtotheserequirements,leavingthestore,orbeingarrested.Atthis
point, Mr. Thornton indicated to the manager that he had always been allowed to
speaktoemployeesonthefloorforasmuchastenminutes,aslongasnobusiness
wasdisrupted,andthathewouldratherbearrestedthanchangetheprocedureofhis
routinevisit.ThemanagerthencalledthepoliceandhadMr.Thorntonhandcuffedin
thestorefortrespassing.Afterhewasbookedandputintoaholdingcellforabrief
time,allchargesweredropped.Mr.ThorntonissuingThriftyDrugforfalsearrest.
Inadditiontothisbackgroundmaterial,whichallparticipantsread,differentgroupswere
exposedtopresentationsbythelawyersforthetwoparties.Naturally,thelawyerforthe
unionorganizerdescribedthearrestasanintimidationattempt,whilethelawyerforthe
store argued that having the talk in the store was disruptive and that the manager was
acting properly. Some participants, like a jury, heard both sides. The lawyers added no
usefulinformationthatyoucouldnotinferfromthebackgroundstory.
The participants were fully aware of the setup, and those who heard only one side
couldeasilyhavegeneratedtheargumentfortheotherside.Nevertheless,thepresentation
of one-sided evidence had a very pronounced effect on judgments. Furthermore,
participants who saw one-sided evidence were more confident of their judgments than
thosewhosawbothsides.Thisisjustwhatyouwouldexpectiftheconfidencethatpeople
experience is determined by the coherence of the story they manage to construct from
availableinformation.Itistheconsistencyoftheinformationthatmattersforagoodstory,
notitscompleteness.Indeed,youwilloftenfindthatknowinglittlemakesiteasiertofit
everythingyouknowintoacoherentpattern.
WY SIATI facilitates the achievement of coherence and of the cognitive ease that
causesustoacceptastatementastrue.Itexplainswhywecanthinkfast,andhowweare
able to make sense of partial information in a complex world. Much of the time, the
coherent story we put together is close enough to reality to support reasonable action.
However,IwillalsoinvokeWYSIATItohelpexplainalonganddiverselistofbiasesof
judgmentandchoice,includingthefollowingamongmanyothers:
Overconfidence:AstheWYSIATIruleimplies,neitherthequantitynorthequality
of the evidence counts for much in subjective confidence. The confidence that
individualshaveintheirbeliefsdependsmostlyonthequalityofthestorytheycan
tell about what they see, even if they see little. We often fail to allow for the
possibilitythatevidencethatshouldbecriticaltoourjudgmentismissing—whatwe
seeisallthereis.Furthermore,ourassociativesystemtendstosettleonacoherent
patternofactivationandsuppressesdoubtandambiguity.
Framing effects: Different ways of presenting the same information often evoke
differentemotions.Thestatementthat“theoddsofsurvivalonemonthaftersurgery
are90%”ismorereassuringthantheequivalentstatementthat“mortalitywithinone
monthofsurgeryis10%.”Similarly,coldcutsdescribedas“90%fat-free”aremore
attractive than when they are described as “10% fat.” The equivalence of the
alternative formulations is transparent, but an individual normally sees only one
formulation,andwhatsheseesisallthereis.
Base-rateneglect:RecallSteve,themeekandtidysoulwhoisoftenbelievedtobea
librarian. The personality description is salient and vivid, and although you surely
knowthattherearemoremalefarmmu
Base-rersthanmalelibrarians, that statisticalfactalmostcertainlydid not cometo
yourmindwhenyoufirstconsideredthequestion.Whatyousawwasalltherewas.
SpeakingofJumpingtoConclusions
“Sheknowsnothingaboutthisperson’smanagementskills.Allsheisgoingbyisthe
haloeffectfromagoodpresentation.”
“Let’s decorrelate errors by obtaining separate judgments on the issue before any
discussion.Wewillgetmoreinformationfromindependentassessments.”
“They made that big decision on the basis of a good report from one consultant.
WYSIATI—what you see is all there is. They did not seem to realize how little
informationtheyhad.”
“Theydidn’twantmoreinformationthatmightspoiltheirstory.WYSIATI.”
P
HowJudgmentsHappen
Thereisnolimittothenumberofquestionsyoucananswer,whethertheyarequestions
someone else asks or questions you ask yourself. Nor is there a limit to the number of
attributesyoucanevaluate.Youarecapableofcountingthenumberofcapitalletterson
thispage,comparingtheheightofthewindowsofyourhousetotheoneacrossthestreet,
and assessing the political prospects of your senator on a scale from excellent to
disastrous.ThequestionsareaddressedtoSystem2,whichwilldirectattentionandsearch
memorytofindtheanswers.System2receivesquestionsorgeneratesthem:ineithercase
it directs attention and searches memory to find the answers. System 1 operates
differently. It continuously monitors what is going on outside and inside the mind, and
continuously generates assessments of various aspects of the situation without specific
intentionand with littleor noeffort.Thesebasicassessments play an important role in
intuitivejudgment,becausetheyareeasilysubstitutedformoredifficultquestions—thisis
theessentialideaoftheheuristicsandbiasesapproach.TwootherfeaturesofSystem1
alsosupportthesubstitutionofonejudgmentforanother.One is theabilitytotranslate
values across dimensions, which you do in answering a question that most people find
easy:“IfSamwereastallasheisintelligent,howtallwouldhebe?”Finally,thereisthe
mental shotgun. An intention of System 2 to answer a specific question or evaluate a
particular attribute of the situation automatically triggers other computations, including
basicassessments.
BasicAssessments
System1hasbeenshapedbyevolutiontoprovideacontinuousassessmentofthemain
problemsthatanorganismmustsolvetosurvive:Howarethingsgoing?Isthereathreat
oramajoropportunity?Iseverythingnormal?ShouldIapproachoravoid?Thequestions
are perhaps less urgent for a human in a city environment than for a gazelle on the
savannah,aalencande:How,butwehaveinheritedtheneuralmechanismsthatevolved
to provide ongoing assessments of threat level, and they have not been turned off.
Situations are constantly evaluated as good or bad, requiring escape or permitting
approach. Good mood and cognitive ease are the human equivalents of assessments of
safetyandfamiliarity.
For a specific example of a basic assessment, consider the ability to discriminate
friendfromfoeataglance.Thiscontributestoone’schancesofsurvivalinadangerous
world,andsuchaspecializedcapabilityhasindeedevolved.AlexTodorov,mycolleague
atPrinceton,hasexploredthebiologicalrootsoftherapidjudgmentsofhowsafeitisto
interactwithastranger.Heshowedthatweareendowedwithanabilitytoevaluate,ina
single glance at a stranger’s face, two potentially crucial facts about that person: how
dominant(andthereforepotentiallythreatening)heis,andhowtrustworthyheis,whether
hisintentionsaremorelikelytobefriendlyorhostile.Theshapeofthefaceprovidesthe
cuesforassessingdominance:a“strong”squarechinisonesuchcue.Facialexpression
(smileorfrown)providesthecuesforassessingthestranger’sintentions.Thecombination
of a square chin with a turned-down mouth may spell trouble. The accuracy of face
reading is far from perfect: round chins are not a reliable indicator of meekness, and
smilescan(tosomeextent)befaked.Still,evenanimperfectabilitytoassessstrangers
confersasurvivaladvantage.
This ancient mechanism is put to a novel use in the modern world: it has some
influence on how people vote. Todorov showed his students pictures of men’s faces,
sometimes for as little as one-tenth of a second, and asked them to rate the faces on
various attributes, including likability and competence. Observers agreed quite well on
those ratings. The faces that Todorov showed were not a random set: they were the
campaignportraitsofpoliticianscompetingforelectiveoffice.Todorovthencomparedthe
results of the electoral races to the ratings of competence that Princeton students had
made,basedonbriefexposuretophotographsandwithoutanypoliticalcontext.Inabout
70% of the races for senator, congressman, and governor, the election winner was the
candidatewhosefacehadearnedahigherratingofcompetence.Thisstrikingresultwas
quicklyconfirmedinnationalelectionsinFinland,inzoningboardelectionsinEngland,
andinvariouselectoralcontestsinAustralia,Germany,andMexico.Surprisingly(atleast
tome),ratingsofcompetencewerefarmorepredictiveofvotingoutcomesinTodorov’s
studythanratingsoflikability.
Todorovhasfoundthatpeoplejudgecompetencebycombiningthetwodimensions
ofstrengthandtrustworthiness.Thefacesthatexudecompetencecombineastrongchin
with a slight confident-appearing smile. There is no evidence that these facial features
actually predict how well politicians will perform in office. But studies of the brain’s
responsetowinningandlosingcandidatesshow thatwearebiologicallypredisposedto
reject candidates who lack the attributes we value—in this research, losers evoked
strongerindicationsof(negative)emotionalresponse.ThisisanexampleofwhatIwill
call a judgment heuristic in the following chapters. Voters are attempting to form an
impression of how good a candidate will be in office, and they fall back on a simpler
assessmentthatismadequicklyandautomaticallyandisavailablewhenSystem2must
makeitsdecision.
PoliticalscientistsfolloweduponTodorov’sinitialresearchbyidentifyingacategory
ofvotersforwhomtheautomaticpreferencesofSystem1areparticularlylikelytoplaya
large role. They found what they were looking for among politicalr m=“5%”>Todoly
uninformedvoterswhowatchagreatdealoftelevision.Asexpected,theeffectoffacial
competence on voting is about three times larger for information-poor and TV-prone
votersthan forothers whoare betterinformed and watch less television. Evidently, the
relative importance of System 1 in determining voting choices is not the same for all
people.Wewillencounterotherexamplesofsuchindividualdifferences.
System1understands language,ofcourse, andunderstandingdepends onthebasic
assessments that are routinely carried out as part of the perception of events and the
comprehension of messages. These assessments include computations of similarity and
representativeness, attributions of causality, and evaluations of the availability of
associationsandexemplars.Theyareperformedevenintheabsenceofaspecifictaskset,
althoughtheresultsareusedtomeettaskdemandsastheyarise.
Thelistofbasicassessmentsislong,butnoteverypossibleattributeisassessed.For
anexample,lookbrieflyatfigure7.
A glance provides an immediate impression of many features of the display. You
knowthatthetwotowersareequallytallandthattheyaremoresimilartoeachotherthan
the tower on the left is to the array of blocks in the middle. However, you do not
immediately know that the number of blocks in the left-hand tower is the same as the
numberofblocksarrayedonthefloor,andyouhavenoimpressionoftheheightofthe
tower that you could build from them. To confirm that the numbers are the same, you
wouldneedtocountthetwosetsofblocksandcomparetheresults,anactivitythatonly
System2cancarryout.
Figure7
SetsandPrototypes
For another example, consider the question: What is the average length of the lines in
figure8?
Figure8
ThisquestioniseasyandSystem1answersitwithoutprompting.Experimentshave
shownthatafractionofasecondissufficientforpeopletoregistertheaveragelengthof
an array of lines with considerable precision. Furthermore, the accuracy of these
judgments is not impaired when the observer is cognitively busy with a memory task.
Theydonotnecessarilyknowhowtodescribetheaverageininchesorcentimeters,but
they will be very accurate in adjusting the length of another line to match the average.
System2isnotneededtoformanimpressionofthenormoflengthforanarray.System1
doesit,automaticallyandeffortlessly,justasitregistersthecolorofthelinesandthefact
thatthey arenot parallel.We also can form an immediateimpressionof thenumber of
objectsinanarray—preciselyiftherearefourorfewerobjects,crudelyiftherearemore.
Nowtoanotherquestion:Whatisthetotallengthofthelinesinfigure8?Thisisa
differentexperience,becauseSystem1hasnosuggestionstooffer.Theonlywayyoucan
answer this question is by activating System 2, which will laboriously estimate the
average,estimateorcountthelines,andmultiplyaveragelengthbythenumberoflines.
estimaight=“0%”>
ThefailureofSystem1tocomputethetotallengthofasetoflinesataglancemay
look obvious to you; you never thought you could do it. It is in fact an instance of an
importantlimitationofthatsystem.BecauseSystem1representscategoriesbyaprototype
orasetoftypicalexemplars,itdealswellwithaveragesbutpoorlywithsums.Thesizeof
thecategory,thenumberofinstancesitcontains,tendstobeignoredinjudgmentsofwhat
Iwillcallsum-likevariables.
Participantsinoneofthenumerousexperimentsthatwerepromptedbythelitigation
followingthedisastrousExxonValdezoilspillwereaskedtheirwillingnesstopayfornets
tocoveroilpondsinwhichmigratorybirdsoftendrown.Differentgroupsofparticipants
statedtheirwillingnesstopaytosave2,000,20,000,or200,000birds.Ifsavingbirdsisan
economic good it should be a sum-like variable: saving 200,000 birds should be worth
muchmorethansaving2,000birds.Infact,theaveragecontributionsofthethreegroups
were $80, $78, and $88 respectively. The number of birds made very little difference.
Whattheparticipantsreactedto,inallthreegroups,wasaprototype—theawfulimageof
ahelplessbirddrowning,itsfeatherssoakedinthickoil.Thealmostcompleteneglectof
quantityinsuchemotionalcontextshasbeenconfirmedmanytimes.
IntensityMatching
Questions about your happiness, the president’s popularity, the proper punishment of
financial evildoers, and the future prospects of a politician share an important
characteristic: they all refer to an underlying dimension of intensity or amount, which
permits the use of the word more: more happy, more popular, more severe, or more
powerful(forapolitician).Forexample,acandidate’spoliticalfuturecanrangefromthe
lowof“Shewillbedefeatedintheprimary”toahighof“Shewillsomedaybepresident
oftheUnitedStates.”
Here we encounter a new aptitude of System 1. An underlying scale of intensity
allows matching across diverse dimensions. If crimes were colors, murder would be a
deepershadeofredthantheft.Ifcrimeswereexpressedasmusic,massmurderwouldbe
playedfortissimowhileaccumulatingunpaidparkingticketswouldbeafaintpianissimo.
And of course you have similar feelings about the intensity of punishments. In classic
experiments, people adjusted the loudness of a sound to the severity of crimes; other
peopleadjustedloudnesstotheseverityoflegalpunishments.Ifyouheardtwonotes,one
forthecrimeandoneforthepunishment,youwouldfeelasenseofinjusticeifonetone
wasmuchlouderthantheother.
Consideranexamplethatwewillencounteragainlater:
Juliereadfluentlywhenshewasfouryearsold.
NowmatchJulie’sreadingprowessasachildtothefollowingintensityscales:
HowtallisamanwhoisastallasJuliewasprecocious?
Whatdoyouthinkof6feet?Obviouslytoolittle.Whatabout7feet?Probablytoomuch.
Youarelookingforaheightthatisasremarkableastheachievementofreadingatage
four. Fairly remarkable, but not extraordinary. Reading at fifteen months would be
extraordinary,perhapslikeamanwhois7′8″.
WhatlevelofincomeinyourprofessionmatchesJulie’sreadingachievement?
WhichcrimeisassevereasJuliewasprecocious?
WhichgraduatingGPAinanIvyLeaguecollegematchesJulie’sreading?
Notveryhard,wasit?Furthermore,youcanbeassuredthatyourmatcheswillbequite
closetothoseofotherpeopleinyourculturalmilieu.Wewillseethatwhenpeopleare
askedtopredictJulie’sGPAfromtheinformationabouttheageatwhichshelearnedto
read,theyanswerbytranslatingfromonescaleto another and pickthematchingGPA.
Andwe willalso see whythis modeof predictionby matching isstatistically wrong—
althoughitisperfectlynaturaltoSystem1,andformostpeopleexceptstatisticiansitis
alsoacceptabletoSystem2.
TheMentalShotgun
System 1 carries out many computations at any one time. Some of these are routine
assessmentsthatgooncontinuously.Wheneveryoureyesareopen,yourbraincomputesa
three-dimensional representation of what is in your field of vision, complete with the
shape of objects, their position in space, and their identity. No intention is needed to
triggerthisoperationorthecontinuousmonitoringforviolatedexpectations.Incontrastto
theseroutineassessments,othercomputationsareundertakenonlywhenneeded:youdo
notmaintainacontinuousevaluationofhowhappyorwealthyyouare,andevenifyou
are a political addict you do not continuously assess the president’s prospects. The
occasionaljudgmentsarevoluntary.Theyoccuronlywhenyouintendthemtodoso.
Youdonotautomaticallycountthenumberofsyllablesofeverywordyouread,but
youcandoit if yousochoose. However, the controloverintendedcomputationsis far
from precise: we often compute much more than we want or need. I call this excess
computationthementalshotgun.Itisimpossibletoaimatasinglepointwithashotgun
becauseitshootspelletsthatscatter,anditseemsalmostequallydifficultforSystem1not
todomorethanSystem2chargesittodo.TwoexperimentsthatIreadlongagosuggested
thisimage.
Participantsinoneexperimentlistenedtopairsofwords,withtheinstructiontopress
akeyas quicklyaspossiblewhenevertheydetectedthatthewordsrhymed.Thewords
rhymeinboththesepairs:
VOTE—NOTE
VOTE—GOAT
Thedifferenceisobvioustoyoubecauseyouseethetwopairs.VOTEandGOATrhyme,
buttheyarespelleddifferently.Theparticipantsonlyheardthewords,buttheywerealso
influencedbythespelling.Theyweredistinctlyslowertorecognizethewordsasrhyming
iftheirspellingwasdiscrepant.Althoughtheinstructionsrequiredonlyacomparisonof
sounds,theparticipantsalsocomparedtheirspelling,andthemismatchontheirrelevant
dimensionslowedthemdown.Anintentiontoansweronequestionevokedanother,which
wasnotonlysuperfluousbutactuallydetrimentaltothemaintask.
Inanotherstudy,peoplelistenedtoaseriesofsentences,withtheinstructiontopress
onekeyasquicklyaspost=“llydesibletoindicateifthesentencewasliterallytrue,and
anotherkeyifthesentencewasnotliterallytrue.Whatarethecorrectresponsesforthe
followingsentences?
Someroadsaresnakes.
Somejobsaresnakes.
Somejobsarejails.
All three sentences are literally false. However, you probably noticed that the second
sentenceismoreobviouslyfalsethantheothertwo—thereactiontimescollectedinthe
experimentconfirmedasubstantialdifference.Thereasonforthedifferenceisthatthetwo
difficult sentences can be metaphorically true. Here again, the intention to perform one
computationevokedanother.Andhereagain,thecorrectanswerprevailedintheconflict,
buttheconflictwiththeirrelevantanswerdisruptedperformance.Inthenextchapterwe
willseethatthecombinationofamentalshotgunwithintensitymatchingexplainswhy
wehaveintuitivejudgmentsaboutmanythingsthatweknowlittleabout.
SpeakingofJudgment
“Evaluating people as attractive or not is a basic assessment. You do that
automaticallywhetherornotyouwantto,anditinfluencesyou.”
“Therearecircuitsinthebrainthatevaluatedominancefromtheshapeoftheface.
Helooksthepartforaleadershiprole.”
“Thepunishmentwon’tfeeljustunlessitsintensitymatchesthecrime.Justlikeyou
canmatchtheloudnessofasoundtothebrightnessofalight.”
“Thiswasaclearinstanceofamentalshotgun.Hewasaskedwhetherhethoughtthe
companywasfinanciallysound,buthecouldn’tforgetthathelikestheirproduct.”
P
AnsweringanEasierQuestion
A remarkable aspect of your mental life is that you are rarely stumped. True, you
occasionallyfaceaquestionsuchas17×24=?towhichnoanswercomesimmediately
tomind,butthesedumbfoundedmomentsarerare.Thenormalstateofyourmindisthat
you have intuitive feelings and opinions about almost everything that comes your way.
Youlikeordislikepeoplelongbeforeyouknowmuchaboutthem;youtrustordistrust
strangerswithoutknowingwhy;youfeelthatanenterpriseisboundtosucceedwithout
analyzingit.Whetheryoustatethemornot,youoftenhaveanswerstoquestionsthatyou
do not completely understand, relying on evidence that you can neither explain nor
defend.
SubstitutingQuestions
Iproposeasimpleaccountofhowwegenerateintuitiveopinionsoncomplexmatters.Ifa
satisfactoryanswertoahardquestionisebrquesDnotfoundquickly,System1willfinda
related question that is easier and will answer it. I call theoperation of answering one
questioninplaceofanothersubstitution.Ialsoadoptthefollowingterms:
Thetargetquestionistheassessmentyouintendtoproduce.
Theheuristicquestionisthesimplerquestionthatyouanswerinstead.
Thetechnicaldefinitionofheuristicisasimpleprocedurethathelpsfindadequate,though
often imperfect, answers to difficult questions. The word comes from the same root as
eureka.
TheideaofsubstitutioncameupearlyinmyworkwithAmos,anditwasthecoreof
whatbecametheheuristicsandbiasesapproach.Weaskedourselveshowpeoplemanage
to make judgments of probability without knowing precisely what probability is. We
concludedthatpeoplemustsomehowsimplifythatimpossibletask,andwesetouttofind
how they do it. Our answer was that when called upon to judge probability, people
actuallyjudge something elseand believethey have judged probability.System 1often
makesthismovewhenfacedwithdifficulttargetquestions,iftheanswertoarelatedand
easierheuristicquestioncomesreadilytomind.
Substituting one question for another can be a good strategy for solving difficult
problems,andGeorgePólyaincludedsubstitutioninhisclassicHowtoSolveIt:“Ifyou
can’t solve a problem, then there is an easier problem you can solve: find it.” Pólya’s
heuristicsarestrategicproceduresthataredeliberatelyimplementedbySystem2.Butthe
heuristics that I discuss in this chapter are not chosen; they are a consequence of the
mentalshotgun,theimprecisecontrolwehaveovertargetingourresponsestoquestions.
Considerthequestionslistedinthe left-hand columnoftable1.Theseare difficult
questions,andbeforeyoucanproduceareasonedanswertoanyofthemyoumustdeal
withotherdifficultissues.Whatisthemeaningofhappiness?Whatarethelikelypolitical
developmentsinthenextsixmonths?Whatarethestandardsentencesforotherfinancial
crimes?Howstrongisthecompetitionthatthecandidatefaces?Whatotherenvironmental
or other causes should be considered? Dealing with these questions seriously is
completely impractical. But you are not limited to perfectly reasoned answers to
questions. There is a heuristic alternative to careful reasoning, which sometimes works
fairlywellandsometimesleadstoseriouserrors.
TargetQuestion HeuristicQuestion
Howmuchwouldyoucontributetosavean
endangeredspecies?
HowmuchemotiondoIfeelwhenI
thinkofdyingdolphins?
Howhappyareyouwithyourlifethese
days? Whatismymoodrightnow?
Howpopularisthepresidentrightnow? Howpopularwillthepresidentbesix
monthsfromnow?
Howshouldfinancialadviserswhopreyonthe
elderlybepunished?
HowmuchangerdoIfeelwhenIthink
offinancialpredators?
Thiswomanisrunningfortheprimary.How
farwillshegoinpolitics?
Doesthiswomanlooklikeapolitical
winner?
Table1
The mental shotgun makes it easy to generate quick answers to difficult questions
withoutimposingmuchhardworkonyourlazySystem2.Theright-handcounterpartof
eachoftheleft-handquestionsisverylikelytobeevokedandveryeasilyanswered.Your
feelingsaboutdolphinsandfinancialcrooks,yourcurrentmood,yourimpressionsofthe
politicalskilloftheprimarycandidate,orthecurrentstandingofthepresidentwillreadily
come to mind. The heuristic questions provide an off-the-shelf answer to each of the
difficulttargetquestions.
Somethingisstillmissingfromthisstory:theanswersneedtobefittedtotheoriginal
questions.Forexample,myfeelingsaboutdyingdolphinsmustbeexpressedindollars.
Another capability of System 1, intensity matching, is available to solve that problem.
Recallthatbothfeelingsandcontributiondollarsareintensityscales.Icanfeelmoreor
lessstronglyaboutdolphinsandthereisacontributionthatmatchestheintensityofmy
feelings.Thedollaramountthatwillcometomymindisthematchingamount.Similar
intensitymatchesarepossibleforallthequestions.Forexample,thepoliticalskillsofa
candidatecanrangefrompathetictoextraordinarilyimpressive,andthescaleofpolitical
successcanrangefromthelowof“Shewillbedefeatedintheprimary”toahighof“She
willsomedaybepresidentoftheUnitedStates.”
The automatic processes of the mental shotgun and intensity matching often make
available one or more answers to easy questions that could be mapped onto the target
question. On some occasions, substitution will occur and a heuristic answer will be
endorsed by System 2. Of course, System 2 has the opportunity to reject this intuitive
answer, or to modify it by incorporating other information. However, a lazy System 2
often follows the path of least effort and endorses a heuristic answer without much
scrutinyofwhetheritistrulyappropriate.Youwillnotbestumped,youwillnothaveto
workveryherрwheard,andyoumaynotevennoticethatyoudidnotanswerthequestion
youwereasked.Furthermore,youmaynotrealizethatthetargetquestionwasdifficult,
becauseanintuitiveanswertoitcamereadilytomind.
The3-DHeuristic
Havealookatthepictureofthethreemenandanswerthequestionthatfollows.
Figure9
Asprintedonthepage,isthefigureontherightlargerthanthefigureontheleft?
Theobviousanswercomesquicklytomind:thefigureontherightislarger.Ifyou
take a ruler to the two figures, however, you will discover that in fact the figures are
exactlythesamesize.Yourimpressionoftheirrelativesizeisdominatedbyapowerful
illusion,whichneatlyillustratestheprocessofsubstitution.
Thecorridorinwhichthefiguresareseenisdrawninperspectiveandappearstogo
into the depth plane. Your perceptual system automatically interprets the picture as a
three-dimensional scene, not as an image printed on a flat paper surface. In the 3-D
interpretation,thepersonontherightisbothmuchfartherawayandmuchlargerthanthe
person on the left. For most of us, this impression of 3-D size is overwhelming. Only
visual artists and experienced photographers have developed the skill of seeing the
drawing as an object on the page. For the rest of us, substitution occurs: the dominant
impression of 3-D size dictates the judgment of 2-D size. The illusion is due to a 3-D
heuristic.
What happens here is a true illusion, not a misunderstanding of the question. You
knewthatthequestionwasaboutthesizeofthefiguresinthepicture,asprintedonthe
page.Ifyouhadbeenaskedtoestimatethesizeofthefigures,weknowfromexperiments
thatyouranswerwouldhavebeenininches,notfeet.Youwerenotconfusedaboutthe
question,butyouwereinfluencedbytheanswertoaquestionthatyouwerenotasked:
“Howtallarethethreepeople?”
The essential step in the heuristic—the substitution of three-dimensional for two-
dimensional size—occurred automatically. The picture contains cues that suggest a 3-D
interpretation.Thesecuesareirrelevanttothetaskathand—thejudgmentofsizeofthe
figure on the page—and you should have ignored them, but you could not. The bias
associatedwiththeheuristicisthatobjectsthatappeartobemoredistantalsoappeartobe
largeron thepage. As thisexample illustrates,a judgment thatis basedon substitution
will inevitably be biased in predictable ways. In this case, it happens so deep in the
perceptualsystemthatyousimplycannothelpit.
TheMoodHeuristicforHappiness
AsurveyofGermanstudentsisoneofthebestexamplesofsubstitution.Thesurveythat
theyoungparticipantscompletedincludedthefollowingtwoquestions:
Howhappyareyouthesedays?
Howmanydatesdidyouhavelastmonth?
< stрr to a p height=“0%” width=“0%”>The experimenters were interested in the
correlation between the two answers. Would the students who reported many dates say
that they were happier than those with fewer dates? Surprisingly, no: the correlation
between the answers was about zero. Evidently, dating was not what came first to the
students’mindswhentheywereaskedtoassesstheirhappiness.Anothergroupofstudents
sawthesametwoquestions,butinreverseorder:
Howmanydatesdidyouhavelastmonth?
Howhappyareyouthesedays?
Theresultsthistimewerecompletelydifferent.Inthissequence,thecorrelationbetween
the number of dates and reported happiness was about as high as correlations between
psychologicalmeasurescanget.Whathappened?
Theexplanationisstraightforward,anditisagoodexampleofsubstitution.Dating
was apparently not the center of these students’ life (in the first survey, happiness and
datingwere uncorrelated),but whenthey wereasked tothink abouttheir romanticlife,
theycertainlyhadanemotionalreaction.Thestudentswhohadmanydateswerereminded
ofahappyaspectoftheirlife,whilethosewhohadnonewereremindedoflonelinessand
rejection.Theemotionarousedbythedatingquestionwasstilloneveryone’smindwhen
thequeryaboutgeneralhappinesscameup.
The psychology of what happened is precisely analogous to the psychology of the
sizeillusioninfigure9.“Happinessthesedays”isnotanaturaloraneasyassessment.A
goodanswerrequiresafairamountofthinking.However,thestudentswhohadjustbeen
askedabouttheirdatingdidnotneedtothinkhardbecausetheyalreadyhadintheirmind
ananswertoarelatedquestion:howhappytheywerewiththeirlovelife.Theysubstituted
thequestiontowhichtheyhadareadymadeanswerforthequestiontheywereasked.
Hereagain,aswedidfortheillusion,wecanask:Arethestudentsconfused?Dothey
reallythinkthatthetwoquestions—theonetheywereaskedandtheonetheyanswer—are
synonymous? Of course not. The students do not temporarily lose their ability to
distinguishromanticlifefromlifeasawhole.Ifaskedaboutthetwoconcepts,theywould
saytheyaredifferent.Buttheywerenotaskedwhethertheconceptsaredifferent.They
wereaskedhowhappytheywere,andSystem1hasareadyanswer.
Dating is not unique. The same pattern is found if a question about the students’
relations with their parents or about their finances immediately precedes the question
about general happiness. In both cases, satisfaction in the particular domain dominates
happinessreports.Anyemotionallysignificantquestionthataltersaperson’smoodwill
havethesameeffect.WYSIATI.Thepresentstateofmindloomsverylargewhenpeople
evaluatetheirhappiness.
TheAffectHeuristic
The dominance of conclusions over arguments is most pronounced where emotions are
involved.ThepsychologistPaulSlovichasproposedanaffectheuristicinwhichpeople
let their likes and dislikes determine their beliefs about the world. Your political
preference determines the arguments that you find compelling. If you like the current
healthpolicy,youbelieveitsbenefitsaresubstantialanditscostsmoremanageablethan
the costs of alternatives. If you are a hawk in your attitude toward other nations, you
probabltheр“0%y think they are relatively weak and likely to submit to your country’s
will.Ifyouareadove,youprobablythinktheyarestrongandwillnotbeeasilycoerced.
Youremotionalattitudetosuchthingsasirradiatedfood,redmeat,nuclearpower,tattoos,
ormotorcyclesdrivesyourbeliefsabouttheirbenefitsandtheirrisks.Ifyoudislikeanyof
thesethings,youprobablybelievethatitsrisksarehighanditsbenefitsnegligible.
Theprimacyofconclusionsdoesnotmeanthatyourmindiscompletelyclosedand
thatyouropinionsarewhollyimmunetoinformationandsensiblereasoning.Yourbeliefs,
andevenyouremotionalattitude,maychange(atleastalittle)whenyoulearnthattherisk
ofanactivityyoudislikedissmallerthanyouthought.However,theinformationabout
lowerriskswillalsochangeyourviewofthebenefits(forthebetter)evenifnothingwas
saidaboutbenefitsintheinformationyoureceived.
We seehereanewsideof the“personality”ofSystem2.UntilnowI have mostly
describeditasamoreorlessacquiescentmonitor,whichallowsconsiderableleewayto
System1.IhavealsopresentedSystem2asactiveindeliberatememorysearch,complex
computations, comparisons, planning, and choice. In the bat-and-ball problem and in
manyotherexamplesoftheinterplaybetweenthetwosystems,itappearedthatSystem2
isultimatelyincharge,withtheabilitytoresistthesuggestionsofSystem1,slowthings
down,andimposelogicalanalysis.Self-criticismisoneofthefunctionsofSystem2.In
thecontextofattitudes,however, System 2 ismoreofan apologistfortheemotionsof
System1thanacriticofthoseemotions—anendorserratherthananenforcer.Itssearch
forinformationandargumentsismostlyconstrainedtoinformationthatisconsistentwith
existing beliefs, not with an intention to examine them. An active, coherence-seeking
System1suggestssolutionstoanundemandingSystem2.
SpeakingofSubstitutionandHeuristics
“Dowestillrememberthequestionwearetryingtoanswer?Orhavewesubstituted
aneasierone?”
“Thequestionwefaceiswhetherthiscandidatecansucceed.Thequestionweseem
toansweriswhethersheinterviewswell.Let’snotsubstitute.”
“Helikesthe project, sohe thinks itscostsare low anditsbenefits arehigh.Nice
exampleoftheaffectheuristic.”
“Weareusinglastyear’sperformanceasaheuristictopredictthevalueofthefirm
severalyearsfromnow.Isthisheuristicgoodenough?Whatotherinformationdowe
need?”
The table below contains a list of features and activities that have been attributed to
System1.Eachoftheactivesentencesreplacesastatement,technicallymoreaccuratebut
hardertounderstand,totheeffectthatamentaleventoccursautomaticallyandfast.My
hopeisthatthelistoftraitswillhelpyoudevelopanintuitivesenseofthe“personality”of
the fictitious System 1. As happens with other characters you know, you will have
hunchesaboutwhatSystem1woulddounderdifferentcircumstances,andmostofyour
huncheswillbecorrect.
CharacteristicsofSystem1
generatesimpressions,feelings,andinclinations;whenendorsedbySystem2these
becomebeliefs,attitudes,andintentions
operatesautomaticallyandquickly,withlittleornoeffort,andnosenseofvoluntary
control
canbeprogrammedbySystem 2 tomobilizeattentionwhenaparticularpattern is
detected(search)
executesskilledresponsesandgeneratesskilledintuitions,afteradequatetraining
createsacoherentpatternofactivatedideasinassociativememory
links a sense of cognitive ease to illusions of truth, pleasant feelings, and reduced
vigilance
distinguishesthesurprisingfromthenormal
infersandinventscausesandintentions
neglectsambiguityandsuppressesdoubt
isbiasedtobelieveandconfirm
exaggeratesemotionalconsistency(haloeffect)
focusesonexistingevidenceandignoresabsentevidence(WYSIATI)
generatesalimitedsetofbasicassessments
representssetsbynormsandprototypes,doesnotintegrate
matchesintensitiesacrossscales(e.g.,sizetoloudness)
computesmorethanintended(mentalshotgun)
sometimessubstitutesaneasierquestionforadifficultone(heuristics)
ismoresensitivetochangesthantostates(prospecttheory)*
overweightslowprobabilities*
showsdiminishingsensitivitytoquantity(psychophysics)*
respondsmorestronglytolossesthantogains(lossaversion)*
framesdecisionproblemsnarrowly,inisolationfromoneanother*
P
Part2
P
HeuristicsandBiases
P
TheLawofSmallNumbers
A study of the incidence of kidney cancer in the 3,141 counties of the United a><
HЉStates reveals a remarkable pattern. The counties in which the incidence of kidney
cancer is lowest are mostly rural, sparsely populated, and located in traditionally
RepublicanstatesintheMidwest,theSouth,andtheWest.Whatdoyoumakeofthis?
Yourmindhasbeenveryactiveinthelastfewseconds,anditwasmainlyaSystem2
operation.Youdeliberatelysearchedmemoryandformulatedhypotheses.Someeffortwas
involved;yourpupilsdilated,andyourheartrateincreasedmeasurably.ButSystem1was
notidle:theoperationofSystem2dependedonthefactsandsuggestionsretrievedfrom
associative memory. You probably rejected the idea that Republican politics provide
protectionagainstkidneycancer.Verylikely,youendedupfocusingonthefactthatthe
counties with low incidence of cancer are mostly rural. The witty statisticians Howard
WainerandHarrisZwerling,fromwhomIlearnedthisexample,commented,“Itisboth
easyandtemptingtoinferthattheirlowcancerratesaredirectlyduetothecleanlivingof
the rural lifestyle—no air pollution, no water pollution, access to fresh food without
additives.”Thismakesperfectsense.
Nowconsiderthecountiesinwhichtheincidenceofkidneycancerishighest.These
ailing counties tend to be mostly rural, sparsely populated, and located in traditionally
RepublicanstatesintheMidwest,theSouth,andtheWest.Tongue-in-cheek,Wainerand
Zwerlingcomment:“Itiseasytoinferthattheirhighcancerratesmightbedirectlydueto
thepovertyoftherurallifestyle—noaccesstogoodmedicalcare,ahigh-fatdiet,andtoo
much alcohol, too much tobacco.” Something is wrong, of course. The rural lifestyle
cannotexplainbothveryhighandverylowincidenceofkidneycancer.
ThekeyfactorisnotthatthecountieswereruralorpredominantlyRepublican.Itis
thatruralcountieshavesmallpopulations.Andthemainlessontobelearnedisnotabout
epidemiology,itisaboutthedifficultrelationshipbetweenourmindandstatistics.System
1 is highly adept in one form of thinking—it automatically and effortlessly identifies
causal connections between events, sometimes even when the connection is spurious.
When told about the high-incidence counties, you immediately assumed that these
counties are different from other counties for a reason, that there must be a cause that
explains this difference. As we shall see, however, System 1 is inept when faced with
“merelystatistical”facts,whichchangetheprobabilityofoutcomesbutdonotcausethem
tohappen.
Arandomevent,bydefinition,doesnotlenditselftoexplanation,butcollectionsof
random events do behave in a highly regular fashion. Imagine a large urn filled with
marbles.Halfthemarblesarered,halfarewhite.Next,imagineaverypatientperson(ora
robot)whoblindlydraws4marblesfromtheurn,recordsthenumberofredballsinthe
sample,throwstheballsbackintotheurn,andthendoesitallagain,manytimes.Ifyou
summarize the results, you will find that the outcome “2 red, 2 white” occurs (almost
exactly) 6 times as often as the outcome “4 red” or “4 white.” This relationship is a
mathematicalfact.Youcanpredicttheoutcomeofrepeatedsamplingfromanurnjustas
confidentlyas you can predict whatwill happenif you hit an eggwith a hammer. You
cannotpredicteverydetailofhowtheshellwillshatter,butyoucanbesureofthegeneral
idea. There is a difference: the satisfying sense of causation that you experience when
thinkingofahammerhittinganeggisaltogetherabsentwhenyouthinkaboutsampling.
Arelatedstatisticalfact is relevanttothe cancer example.Fromthe same urn,two
verypatient marblecounters thatрy dake turns. Jackdraws 4 marbleson each trial, Jill
draws7.Theybothrecordeachtimetheyobserveahomogeneoussample—allwhiteorall
red.Iftheygoonlongenough,Jackwillobservesuchextremeoutcomesmoreoftenthan
Jill—byafactorof8(theexpectedpercentagesare12.5%and1.56%).Again,nohammer,
nocausation,butamathematicalfact:samples of 4marblesyieldextremeresultsmore
oftenthansamplesof7marblesdo.
Now imagine the population of the United States as marbles in a giant urn. Some
marblesare markedKC, forkidney cancer.You drawsamples ofmarbles andpopulate
eachcountyinturn.Ruralsamplesaresmallerthanothersamples.Justasinthegameof
JackandJill,extremeoutcomes(veryhighand/orverylowcancerrates)aremostlikelyto
befoundinsparselypopulatedcounties.Thisisallthereistothestory.
We started from a fact that calls for a cause: the incidence of kidney cancer varies
widely across counties and the differences are systematic. The explanation I offered is
statistical:extremeoutcomes(bothhighandlow)aremorelikelytobefoundinsmallthan
inlargesamples.Thisexplanationisnotcausal.Thesmallpopulationofacountyneither
causesnorpreventscancer;itmerelyallowstheincidenceofcancertobemuchhigher(or
muchlower)thanitisinthelargerpopulation.Thedeepertruthisthatthereisnothingto
explain.Theincidenceofcancerisnottrulylowerorhigherthannormalinacountywith
asmallpopulation,itjustappearstobesoinaparticularyearbecauseofanaccidentof
sampling.Ifwerepeattheanalysisnextyear,wewillobservethesamegeneralpatternof
extremeresultsinthesmallsamples,butthecountieswherecancerwascommonlastyear
will not necessarily have a high incidence this year. If this is the case, the differences
betweendenseandruralcountiesdonotreallycountasfacts:theyarewhatscientistscall
artifacts,observationsthatareproducedentirelybysomeaspectofthemethodofresearch
—inthiscase,bydifferencesinsamplesize.
ThestoryIhavetoldmayhavesurprisedyou,butitwasnotarevelation.Youhave
longknownthattheresultsoflargesamplesdeservemoretrustthansmallersamples,and
evenpeoplewhoareinnocentofstatisticalknowledgehaveheardaboutthislawoflarge
numbers. But “knowing” is not a yes-no affair and you may find that the following
statementsapplytoyou:
Thefeature“sparselypopulated”didnotimmediatelystandoutasrelevantwhenyou
readtheepidemiologicalstory.
Youwereatleastmildlysurprisedbythesizeofthedifferencebetweensamplesof4
andsamplesof7.
Evennow,youmustexertsomementalefforttoseethatthefollowingtwostatements
meanexactlythesamething:
Largesamplesaremoreprecisethansmallsamples.
Smallsamplesyieldextremeresultsmoreoftenthanlargesamplesdo.
Thefirststatementhasaclearringoftruth,butuntilthesecondversionmakesintuitive
sense,youhavenottrulyunderstoodthefirst.
Thebottomline:yes,youdidknowthattheresultsoflargesamplesaremoreprecise,
butyoumaynowrealizethatyoudidnotknowitverywell.Youarenotalone.Thefirst
studythatAmosandIdidtogethershowedthatevensophisticatedresearchershavepoor
intuitionsandawobblyunderstandingofsamplingeffects.
TheLawofSmallNumbers
MycollaborationwithAmosintheearly1970sbeganwithadiscussionoftheclaimthat
peoplewhohavehadnotraininginstatisticsaregood“intuitivestatisticians.”Hetoldmy
seminar and me of researchers at the University of Michigan who were generally
optimisticabout intuitivestatistics.I had strong feelings about that claim, which I took
personally:IhadrecentlydiscoveredthatIwasnotagoodintuitivestatistician,andIdid
notbelievethatIwasworsethanothers.
Foraresearchpsychologist,samplingvariationisnotacuriosity;itisanuisanceand
a costly obstacle, which turns the undertaking of every research project into a gamble.
Supposethatyouwishtoconfirmthehypothesisthatthevocabularyoftheaveragesix-
year-old girl is larger than the vocabulary of an average boy of the same age. The
hypothesisistrueinthepopulation;theaveragevocabularyofgirlsisindeedlarger.Girls
andboysvaryagreatdeal,however,andbytheluckofthedrawyoucouldselectasample
inwhichthedifferenceisinconclusive,orevenoneinwhichboysactuallyscorehigher.If
youaretheresearcher,thisoutcomeiscostlytoyoubecauseyouhavewastedtimeand
effort,andfailedtoconfirmahypothesisthatwasinfacttrue.Usingasufficientlylarge
sampleistheonlywaytoreducetherisk.Researcherswhopicktoosmallasampleleave
themselvesatthemercyofsamplingluck.
The risk of error can be estimated for any given sample size by a fairly simple
procedure. Traditionally, however, psychologists do not use calculations to decide on a
samplesize.Theyusetheirjudgment,whichiscommonlyflawed.AnarticleIhadread
shortlybeforethedebatewithAmosdemonstratedthemistakethatresearchersmade(they
stilldo)byadramaticobservation.Theauthorpointedoutthatpsychologistscommonly
chosesamplessosmallthattheyexposedthemselvestoa50%riskoffailingtoconfirm
their true hypotheses! No researcher in his right mind would accept such a risk. A
plausible explanation was that psychologists’ decisions about sample size reflected
prevalentintuitivemisconceptionsoftheextentofsamplingvariation.
The article shocked me, because it explained some troubles I had had in my own
research.Likemostresearchpsychologists,Ihadroutinelychosensamplesthatweretoo
smallandhadoftenobtainedresultsthatmadenosense.NowIknewwhy:theoddresults
wereactuallyartifactsofmyresearchmethod.Mymistakewasparticularlyembarrassing
becauseItaughtstatisticsandknewhowtocomputethesamplesizethatwouldreducethe
risk of failure to an acceptable level. But I had never chosen a sample size by
computation.Likemycolleagues,Ihadtrustedtraditionandmyintuitioninplanningmy
experiments and had never thought seriously about the issue. When Amos visited the
seminar,Ihadalreadyreachedtheconclusionthatmyintuitionsweredeficient,andinthe
courseoftheseminarwequicklyagreedthattheMichiganoptimistswerewrong.
AmosandIsetouttoexaminewhetherIwastheonlyfooloramemberofamajority
offools,bytestingwhetherresearchersselectedformathematicalexpertisewouldmake
similarmistakes.Wedevelopedaquestionnairethatdescribedrealisticresearchsituations,
includingreplicationsofsuccessfulexperiments.Itaskedtheresearcherstochoosesample
sizes,toassesstherisksoffailuretowhichtheirdecisionsexposedthem,andtoprovide
advice to hypothetical graduate students planning their research. Amos collected the
responses of a group of sophisticated participants (including authors of two statistical
textbooks)atameetatiрp>
AmosandIcalledourfirstjointarticle“BeliefintheLawofSmallNumbers.”We
explained,tongue-in-cheek,that“intuitionsaboutrandomsamplingappeartosatisfythe
law of small numbers, which asserts that the law of large numbers applies to small
numbersaswell.”Wealsoincludedastronglywordedrecommendationthatresearchers
regardtheir“statisticalintuitionswithpropersuspicionandreplaceimpressionformation
bycomputationwheneverpossible.”
ABiasofConfidenceOverDoubt
Inatelephonepollof300seniors,60%supportthepresident.
Ifyouhadtosummarizethemessageofthissentenceinexactlythreewords,whatwould
they be? Almost certainly you would choose “elderly support president.” These words
providethegistofthestory.Theomitteddetailsofthepoll,thatitwasdoneonthephone
with a sample of 300, are of no interest in themselves; they provide background
informationthatattractslittleattention.Yoursummarywouldbethesameifthesample
sizehadbeendifferent.Ofcourse,acompletelyabsurdnumberwoulddrawyourattention
(“atelephonepollof6[or60million]elderlyvoters…”).Unlessyouareaprofessional,
however,youmaynotreactverydifferentlytoasampleof150andtoasampleof3,000.
Thatisthemeaningofthestatementthat“peoplearenotadequatelysensitivetosample
size.”
The message about the poll contains information of two kinds: the story and the
sourceofthestory.Naturally,youfocusonthestoryratherthanonthereliabilityofthe
results.Whenthereliabilityisobviouslylow,however,themessagewillbediscredited.If
youaretoldthat“apartisangrouphasconductedaflawedandbiasedpolltoshowthatthe
elderlysupportthepresident…”youwillofcourserejectthefindingsofthepoll,andthey
willnotbecomepartofwhatyoubelieve.Instead,thepartisanpollanditsfalseresults
willbecomeanewstoryaboutpoliticallies.Youcanchoosetodisbelieveamessagein
suchclear-cutcases.Butdoyoudiscriminatesufficientlybetween“IreadinTheNewYork
Times…”and“Iheardatthewatercooler…”?CanyourSystem1distinguishdegreesof
belief?TheprincipleofWYSIATIsuggeststhatitcannot.
AsIdescribedearlier,System1isnotproneto doubt. Itsuppressesambiguityand
spontaneously constructs stories that are as coherent as possible. Unless the message is
immediately negated, the associations that it evokes will spread as if the message were
true.System2iscapableofdoubt,becauseitcanmaintainincompatiblepossibilitiesat
thesametime.However,sustainingdoubtisharderworkthanslidingintocertainty.The
lawofsmallnumbersisamanifestationofageneralbiasthatfavorscertaintyoverdoubt,
whichwillturnupinmanyguisesinfollowingchapters.
Thestrongbiastowardbelievingthatsmallsamplescloselyresemblethepopulation
fromwhichtheyaredrawnisalsopartofalargerstory:wearepronetoexaggeratethe
consistencyandcoherenceofwhatwesee.Theexaggeratedfaithofresearchersinwhat
can be learned from a few observations is closely related to the halo effect thрhe , the
senseweoftengetthatweknowandunderstandapersonaboutwhomweactuallyknow
verylittle.System1runsaheadofthefactsinconstructingarichimageonthebasisof
scrapsofevidence.Amachineforjumpingtoconclusionswillactasifitbelievedinthe
law of small numbers. More generally, it will produce a representation of reality that
makestoomuchsense.
CauseandChance
Theassociativemachineryseekscauses.Thedifficultywehavewithstatisticalregularities
is that they call for a different approach. Instead of focusing on how the event at hand
cametobe,thestatisticalviewrelatesittowhatcouldhavehappenedinstead.Nothingin
particularcausedittobewhatitis—chanceselecteditfromamongitsalternatives.
Ourpredilectionforcausalthinkingexposesustoseriousmistakesinevaluatingthe
randomnessof truly randomevents. For anexample, take thesex of sixbabies bornin
sequenceatahospital.Thesequenceofboysandgirlsisobviouslyrandom;theeventsare
independentofeachother,andthenumberofboysandgirlswhowereborninthehospital
inthelastfewhourshasnoeffectwhatsoeveronthesexofthenextbaby.Nowconsider
threepossiblesequences:
BBBGGG
GGGGGG
BGBBGB
Are the sequences equally likely? The intuitive answer—“of course not!”—is false.
Because the events are independent and because the outcomes B and G are
(approximately)equallylikely,thenanypossiblesequenceofsixbirthsisaslikelyasany
other.Evennowthatyouknowthisconclusionistrue,itremainscounterintuitive,because
only the third sequence appears random. As expected, BGBBGB is judged much more
likelythantheothertwosequences.Wearepatternseekers,believersinacoherentworld,
inwhichregularities(suchasasequenceofsixgirls)appearnotbyaccidentbutasaresult
of mechanical causality or of someone’s intention. We do not expect to see regularity
producedbyarandomprocess,andwhenwedetectwhatappearstobearule,wequickly
reject the idea that the process is truly random. Random processes produce many
sequencesthatconvincepeoplethattheprocessisnotrandomafterall.Youcanseewhy
assuming causality could have had evolutionary advantages. It is part of the general
vigilancethatwehaveinheritedfromancestors.Weareautomaticallyonthelookoutfor
thepossibilitythattheenvironmenthaschanged.Lionsmayappearontheplainatrandom
times,butitwouldbesafertonoticeandrespondtoanapparentincreaseintherateof
appearance of prides of lions, even if it is actually due to the fluctuations of a random
process.
The widespread misunderstanding of randomness sometimes has significant
consequences. In our article on representativeness, Amos and I cited the statistician
WilliamFeller,whoillustratedtheeasewithwhichpeopleseepatternswherenoneexists.
DuringtheintensiverocketbombingofLondoninWorldWarII,itwasgenerallybelieved
that the bombing could not be random because a map of the hits revealed conspicuous
gaps.SomesuspectedthatGerman spies were locatedintheunharmedareas. A careful
statisticalanalysisrevealedthatthedistributionofhitswastypicalofarandomprocess—
and typical as well in evoking a strong impression that it was not random. “To the
untrainedeye,”Fellerremarks,“randomnessappearsasregularityortendencytocluster.”
I soon had an occasion to apply what I had learned frpeaрrainom Feller. The Yom
KippurWarbrokeoutin1973,andmyonlysignificantcontributiontothewareffortwas
toadvisehighofficersintheIsraeliAirForcetostopaninvestigation.Theairwarinitially
wentquite badly for Israel, becauseof theunexpectedly good performanceof Egyptian
ground-to-airmissiles.Losseswerehigh,andtheyappearedtobeunevenlydistributed.I
wastoldoftwosquadronsflyingfromthesamebase,oneofwhichhadlostfourplanes
whiletheotherhadlostnone.Aninquirywasinitiatedinthehopeoflearningwhatitwas
thattheunfortunatesquadronwasdoingwrong.Therewasnopriorreasontobelievethat
one of the squadrons was more effective than the other, and no operational differences
werefound,butofcoursethelivesofthepilotsdifferedinmanyrandomways,including,
asIrecall,howoftentheywenthomebetweenmissionsandsomethingabouttheconduct
ofdebriefings.Myadvicewasthatthecommandshouldacceptthatthedifferentoutcomes
wereduetoblindluck,andthattheinterviewingofthepilotsshouldstop.Ireasonedthat
luck was the most likely answer, that a random search for a nonobvious cause was
hopeless,andthatinthemeantimethepilotsinthesquadronthathadsustainedlossesdid
notneedtheextraburdenofbeingmadetofeelthattheyandtheirdeadfriendswereat
fault.
Someyearslater,AmosandhisstudentsTomGilovichandRobertVallonecauseda
stirwiththeirstudyofmisperceptionsofrandomnessinbasketball.The“fact”thatplayers
occasionallyacquireahothandisgenerallyacceptedbyplayers,coaches,andfans.The
inferenceisirresistible:aplayersinksthreeorfourbasketsinarowandyoucannothelp
forming the causal judgment that this player is now hot, with a temporarily increased
propensitytoscore.Playersonbothteamsadapttothisjudgment—teammatesaremore
likelytopasstothehotscorerandthedefenseismorelikelytodoubleteam.Analysisof
thousandsofsequencesofshotsledtoadisappointingconclusion:thereisnosuchthing
asahothandinprofessionalbasketball,eitherinshootingfromthefieldorscoringfrom
thefoulline.Ofcourse,someplayersaremoreaccuratethanothers,butthesequenceof
successesandmissedshotssatisfiesalltestsofrandomness.Thehothandisentirelyinthe
eye of the beholders, who are consistently too quick to perceive order and causality in
randomness.Thehothandisamassiveandwidespreadcognitiveillusion.
Thepublicreactiontothisresearchispartofthestory.Thefindingwaspickedupby
the press because of its surprising conclusion, and the general response was disbelief.
WhenthecelebratedcoachoftheBostonCeltics,RedAuerbach,heardofGilovichand
hisstudy,heresponded,“Whoisthisguy?Sohemakesastudy.Icouldn’tcareless.”The
tendencytoseepatternsinrandomnessisoverwhelming—certainlymoreimpressivethan
aguymakingastudy.
Theillusionofpatternaffectsourlivesinmanywaysoffthebasketballcourt.How
many good years should you wait before concluding that an investment adviser is
unusually skilled? How many successful acquisitions should be needed for a board of
directorstobelievethattheCEOhasextraordinaryflairforsuchdeals?Thesimpleanswer
tothesequestionsisthatifyoufollowyourintuition,youwillmoreoftenthannoterrby
misclassifyingarandomeventassystematic.Wearefartoowillingtorejectthebeliefthat
muchofwhatweseeinlifeisrandom.
IbeganthischapterwiththeexampleofcancerincidenceacrosstheUnitedStates.
Theexampleappearsinabookintendedforstatisticsteachers,butIlearnedaboutitfrom
an amusing article by the two statisticians I quoted earlier, Howard Wainer and Harris
Zwerling.Theiressayfocusedonalargeiiveрothersnvestment,some$1.7billion,which
theGatesFoundationmadetofollowupintriguingfindingsonthecharacteristicsofthe
mostsuccessfulschools.Manyresearchershavesoughtthesecretofsuccessfuleducation
byidentifyingthemostsuccessfulschoolsinthehopeofdiscoveringwhatdistinguishes
them from others. One of the conclusions of this research is that the most successful
schools,onaverage,aresmall.Inasurveyof1,662schoolsinPennsylvania,forinstance,
6ofthetop50weresmall,whichis an overrepresentationby afactorof4.Thesedata
encouragedtheGatesFoundationtomakeasubstantialinvestmentinthecreationofsmall
schools,sometimesbysplittinglargeschoolsintosmallerunits.Atleasthalfadozenother
prominentinstitutions,suchastheAnnenbergFoundationandthePewCharitableTrust,
joined the effort, as did the U.S. Department of Education’s Smaller Learning
CommunitiesProgram.
Thisprobablymakesintuitivesensetoyou.Itiseasytoconstructacausalstorythat
explainshowsmallschoolsareabletoprovidesuperioreducationandthusproducehigh-
achievingscholarsbygivingthemmorepersonalattentionandencouragementthanthey
couldgetinlargerschools.Unfortunately,thecausalanalysisispointlessbecausethefacts
arewrong.IfthestatisticianswhoreportedtotheGatesFoundationhadaskedaboutthe
characteristicsoftheworstschools,theywouldhavefoundthatbadschoolsalsotendto
besmallerthanaverage.Thetruthisthatsmallschoolsarenotbetteronaverage;theyare
simplymorevariable.Ifanything,sayWainerandZwerling,largeschoolstendtoproduce
betterresults,especiallyinhighergradeswhereavarietyofcurricularoptionsisvaluable.
Thanks to recent advances in cognitive psychology, we can now see clearly what
Amos and I could only glimpse: the law of small numbers is part of two larger stories
abouttheworkingsofthemind.
The exaggerated faith in small samples is only one example of a more general
illusion—wepaymoreattentiontothecontentofmessagesthantoinformationabout
their reliability, and as a result end up with a view of the world around us that is
simpler and more coherent than the data justify. Jumping to conclusions is a safer
sportintheworldofourimaginationthanitisinreality.
Statisticsproducemanyobservationsthatappeartobegforcausalexplanationsbut
do not lend themselves to such explanations. Many facts of the world are due to
chance, including accidents of sampling. Causal explanations of chance events are
inevitablywrong.
SpeakingoftheLawofSmallNumbers
“Yes,thestudiohashadthreesuccessfulfilmssincethenewCEOtookover.Butitis
tooearlytodeclarehehasahothand.”
“Iwon’tbelievethatthenewtraderisageniusbeforeconsultingastatisticianwho
couldestimatethelikelihoodofhisstreakbeingachanceevent.”
“Thesampleofobservationsistoosmalltomakeanyinferences.Let’snotfollowthe
lawofsmallnumbers.”
“Iplantokeeptheresultsoftheexperimentsecretuntilwehaveasufficientlylarge
sample.Otherwisortрxperewewillfacepressuretoreachaconclusionprematurely.”
P
Anchors
AmosandIonceriggedawheeloffortune.Itwasmarkedfrom0to100,butwehadit
built so that it would stop only at 10 or 65. We recruited students of the University of
Oregonasparticipantsinourexperiment.Oneofuswouldstandinfrontofasmallgroup,
spin the wheel, and ask them to write down the number on which the wheel stopped,
whichofcoursewaseither10or65.Wethenaskedthemtwoquestions:
IsthepercentageofAfricannationsamongUNmemberslargerorsmallerthanthe
numberyoujustwrote?
WhatisyourbestguessofthepercentageofAfricannationsintheUN?
Thespinofawheeloffortune—evenonethatisnotrigged—cannotpossiblyyielduseful
information about anything, and the participants in our experiment should simply have
ignoredit.Buttheydidnotignoreit.Theaverageestimatesofthosewhosaw10and65
were25%and45%,respectively.
Thephenomenonwewerestudyingissocommonandsoimportantintheeveryday
world that you should know its name: it is an anchoring effect. It occurs when people
consideraparticularvalueforanunknownquantitybeforeestimatingthatquantity.What
happens is one of the most reliable and robust results of experimental psychology: the
estimatesstayclosetothenumberthatpeopleconsidered—hencetheimageofananchor.
IfyouareaskedwhetherGandhiwasmorethan114yearsoldwhenhediedyouwillend
up with a much higher estimate of his age at death than you would if the anchoring
questionreferredtodeathat35.Ifyouconsiderhowmuchyoushouldpayforahouse,
youwillbeinfluencedbytheaskingprice.Thesamehousewillappearmorevaluableif
itslistingpriceishighthanifitislow,evenifyouaredeterminedtoresisttheinfluenceof
thisnumber;andsoon—thelistofanchoringeffectsisendless.Anynumberthatyouare
askedtoconsiderasapossiblesolutiontoanestimationproblemwillinduceananchoring
effect.
Wewerenotthefirsttoobservetheeffectsofanchors,butourexperimentwasthe
firstdemonstrationofitsabsurdity:people’sjudgmentswereinfluencedbyanobviously
uninformativenumber.Therewasnowaytodescribetheanchoringeffectofawheelof
fortuneasreasonable.AmosandIpublishedtheexperimentinourSciencepaper,anditis
oneofthebestknownofthefindingswereportedthere.
Therewasonlyonetrouble:AmosandIdidnotfullyagreeonthepsychologyofthe
anchoringeffect.Hesupportedoneinterpretation,Ilikedanother,andweneverfounda
waytosettletheargument.Theproblemwasfinallysolveddecadeslaterbytheeffortsof
numerous investigators. It is now clear that Amos and I were both right. Two different
mechanisms produce anchoring effects—one for each system. There is a form of
anchoringthatoccursinadeliberateprocessofadjustment,anoperationofSystem2.And
thereisanchoringthatoccursbyaprimingeffect,anautomaticmanifestationofSystem1.
AnchoringasAdjustment
Amos liked the idea of an adjust-and-anchor heuristic as a strategy for estimating
uncertainquantities:startfromananchoringnumber,assesswhetheritistoohighortoo
low, and gradually adjust your estimate by mentally “moving” from the anchor. The
adjustment typically ends prematurely, because people stop when they are no longer
certain that they should move farther. Decades after our disagreement, and years after
Amos’sdeath,convincingevidenceofsuchaprocesswasofferedindependentlybytwo
psychologistswhohadworkedcloselywithAmosearlyintheircareers:EldarShafirand
TomGilovichtogetherwiththeirownstudents—Amos’sintellectualgrandchildren!
Togettheidea,takeasheetofpaperanddrawa2½-inchlinegoingup,startingatthe
bottomofthepage—withoutaruler.Nowtakeanothersheet,andstartatthetopanddraw
alinegoingdown untilitis2½ inches fromthebottom.Compare the lines.Thereisa
goodchancethatyourfirstestimateof2½incheswasshorterthanthesecond.Thereason
isthatyoudonotknowexactlywhatsuchalinelookslike;thereisarangeofuncertainty.
Youstopnearthebottomoftheregionofuncertaintywhenyoustartfromthebottomof
thepageandnearthetopoftheregionwhenyoustartfromthetop.RobynLeBoeufand
Shafir found many examples of that mechanism in daily experience. Insufficient
adjustment neatly explains why you are likely to drive too fast when you come off the
highway onto city streets—especially if you are talking with someone as you drive.
Insufficient adjustment is also a source of tension between exasperated parents and
teenagers who enjoy loud music in their room. Le Boeuf and Shafir note that a “well-
intentionedchildwhoturnsdownexceptionallyloudmusictomeetaparent’sdemandthat
itbeplayedata‘reasonable’volumemayfailtoadjustsufficientlyfromahighanchor,
andmayfeelthatgenuineattemptsatcompromisearebeingoverlooked.”Thedriverand
thechildbothdeliberatelyadjustdown,andbothfailtoadjustenough.
Nowconsiderthesequestions:
WhendidGeorgeWashingtonbecomepresident?
WhatistheboilingtemperatureofwateratthetopofMountEverest?
Thefirstthingthathappenswhenyouconsidereachofthesequestionsisthatananchor
comestoyourmind,andyouknowboththatitiswrongandthedirectionofthecorrect
answer.YouknowimmediatelythatGeorgeWashingtonbecamepresidentafter1776,and
youalsoknowthattheboilingtemperatureofwateratthetopofMountEverestislower
than100°C.Youhavetoadjustintheappropriatedirectionbyfindingargumentstomove
awayfromtheanchor.Asinthecaseofthelines,youarelikelytostopwhenyouareno
longersureyoushouldgofarther—atthenearedgeoftheregionofuncertainty.
NickEpleyandTomGilovichfoundevidencethatadjustmentisadeliberateattemptto
findreasonstomoveawayfromtheanchor:peoplewhoareinstructedtoshaketheirhead
whentheyheartheanchor,asiftheyrejectedit,movefartherfromtheanchor,andpeople
who nod their head show enhanced anchoring. Epley and Gilovich also confirmed that
adjustmentisan effortful operation.People adjust less(staycloser tothe anchor) when
their mental resources are depleted, either because their memory is loaded with
dighdth=igitsorbecausetheyareslightlydrunk.Insufficientadjustmentisafailureofa
weakorlazySystem2.
Sowenowknow thatAmoswasrightfor at leastsomecasesof anchoring, which
involveadeliberateSystem2adjustmentinaspecifieddirectionfromananchor.
AnchoringasPrimingEffect
WhenAmosandIdebated anchoring, Iagreedthatadjustmentsometimesoccurs,but I
was uneasy. Adjustment is a deliberate and conscious activity, but in most cases of
anchoringthereisnocorrespondingsubjectiveexperience.Considerthesetwoquestions:
WasGandhimoreorlessthan144yearsoldwhenhedied?
HowoldwasGandhiwhenhedied?
Did you produce your estimate by adjusting down from 144? Probably not, but the
absurdlyhighnumberstillaffectedyourestimate.Myhunchwasthatanchoringisacase
of suggestion. This is the word we use when someone causes us to see, hear, or feel
somethingbymerelybringingittomind.Forexample,thequestion“Doyounowfeela
slightnumbnessinyourleftleg?”alwayspromptsquiteafewpeopletoreportthattheir
leftlegdoesindeedfeelalittlestrange.
AmoswasmoreconservativethanIwasabouthunches,andhecorrectlypointedout
that appealing to suggestion did not help us understand anchoring, because we did not
know how to explain suggestion. I had to agree that he was right, but I never became
enthusiastic about the idea of insufficient adjustment as the sole cause of anchoring
effects. We conducted many inconclusive experiments in an effort to understand
anchoring,butwefailedandeventuallygaveuptheideaofwritingmoreaboutit.
Thepuzzlethatdefeatedusisnowsolved,becausetheconceptofsuggestionisno
longer obscure: suggestion is a priming effect, which selectively evokes compatible
evidence. You did not believe for a moment that Gandhi lived for 144 years, but your
associativemachinerysurelygeneratedanimpressionofaveryancientperson.System1
understands sentences by trying to make them true, and the selective activation of
compatible thoughts produces a family of systematic errors that make us gullible and
pronetobelievetoostronglywhateverwebelieve.WecannowseewhyAmosandIdid
notrealizethatthereweretwotypesofanchoring:theresearchtechniquesandtheoretical
ideasweneededdidnotyetexist.Theyweredeveloped,muchlater,byotherpeople.A
processthatresemblessuggestionisindeedatworkinmanysituations:System1triesits
best to construct a world in which the anchor is the true number. This is one of the
manifestationsofassociativecoherencethatIdescribedinthefirstpartofthebook.
The German psychologists Thomas Mussweiler and Fritz Strack offered the most
compelling demonstrations of the role of associative coherence in anchoring. In one
experiment, they asked an anchoring question about temperature: “Is the annual mean
temperature in Germany higher or lower than 20°C (68°F)?” or “Is the annual mean
temperatureinGermanyhigherorlowerthan5°C(40°F)?”
Allparticipantswerethenbrieflyshownwordsthattheywereaskedtoidentify.The
researchers found that 68°F made it easier to recognize summer words (like sun and
beach),and40°Ffacilitatedwinterwords(likefrostandski).Theselectiveactivationof
compatiblememoriesexplainsanchoring:thehighandthelownumbersactivatedifferent
sets of ideas in memory. The estimates of annual temperature draw on these biased
samplesof ideasand aretherefore biasedas well.In anotherelegant studyin thesame
vein, participants were asked about the average price of German cars. A high anchor
selectivelyprimedthenamesofluxurybrands(Mercedes,Audi),whereasthelowanchor
primedbrandsassociated withmass-marketcars (Volkswagen). We sawearlier that any
primewilltendtoevokeinformationthatiscompatiblewithit.Suggestionandanchoring
arebothexplainedbythesameautomaticoperationofSystem1.AlthoughIdidnotknow
howtoproveitatthetime,myhunchaboutthelinkbetweenanchoringandsuggestion
turnedouttobecorrect.
TheAnchoringIndex
Manypsychologicalphenomenacanbedemonstratedexperimentally,butfewcanactually
bemeasured.Theeffectofanchorsisanexception.Anchoringcanbemeasured,anditis
animpressivelylargeeffect.SomevisitorsattheSanFranciscoExploratoriumwereasked
thefollowingtwoquestions:
Istheheightofthetallestredwoodmoreorlessthan1,200feet?
Whatisyourbestguessabouttheheightofthetallestredwood?
The “high anchor” in this experiment was 1,200 feet. For other participants, the first
questionreferredtoa“lowanchor”of180feet.Thedifferencebetweenthetwoanchors
was1,020feet.
As expected, the two groups produced very different mean estimates: 844 and 282
feet.Thedifferencebetweenthemwas562feet.Theanchoringindexissimplytheratioof
thetwodifferences(562/1,020)expressedasapercentage:55%.Theanchoringmeasure
wouldbe 100% forpeople who slavishlyadopt the anchoras anestimate,and zerofor
peoplewhoareabletoignoretheanchoraltogether.Thevalueof55%thatwasobserved
inthisexampleistypical.Similarvalueshavebeenobservedinnumerousotherproblems.
Theanchoringeffectisnotalaboratorycuriosity;itcanbejustasstronginthereal
world. In an experiment conducted some years ago, real-estate agents were given an
opportunitytoassessthevalueofahousethatwasactuallyonthemarket.Theyvisitedthe
houseandstudiedacomprehensivebookletofinformationthatincludedanaskingprice.
Halftheagentssawanaskingpricethatwassubstantiallyhigherthanthelistedpriceof
thehouse;theotherhalfsawanaskingpricethatwassubstantiallylower.Eachagentgave
heropinionaboutareasonablebuyingpriceforthehouseandthelowestpriceatwhich
shewouldagreetosellthehouseifsheownedit.Theagentswerethenaskedaboutthe
factorsthathadaffectedtheirjudgment.Remarkably,theaskingpricewasnotoneofthese
factors;theagentstookprideintheirabilitytoignoreit.Theyinsistedthatthelistingprice
had no effect on their responses, but they were wrong: the anchoring effect was 41%.
Indeed, the professionals were almost as susceptible to anchoring effects as business
schoolstudentswithnoreal-estateexperience,whoseanchoringindexwas48%.Theonly
difference between the two groups was that the students conceded that they were
influencedbytheanchor,whiletheprofessionalsdeniedthatinfluence.
Powerful anchoring effects are found in decisions that people make about money,
suchaswhentheychoosehowmuchtocontributeal.lsdeniedtoacause.Todemonstrate
this effect, we told participants in the Exploratorium study about the environmental
damagecausedbyoiltankersinthePacificOceanand asked abouttheirwillingnessto
makeanannualcontribution“tosave50,000offshorePacificCoastseabirdsfromsmall
offshoreoilspills,untilwaysarefoundtopreventspillsorrequiretankerownerstopay
fortheoperation.”Thisquestionrequiresintensitymatching:therespondentsareasked,in
effect, to find the dollar amount of a contribution that matches the intensity of their
feelingsabouttheplightoftheseabirds.Someofthevisitorswerefirstaskedananchoring
question,suchas,“Wouldyoubewillingtopay$5…,”beforethepoint-blankquestionof
howmuchtheywouldcontribute.
When no anchor was mentioned, the visitors at the Exploratorium—generally an
environmentallysensitivecrowd—saidtheywerewillingtopay$64,onaverage.When
theanchoringamountwasonly$5,contributionsaveraged$20.Whentheanchorwasa
ratherextravagant$400,thewillingnesstopayrosetoanaverageof$143.
The difference between the high-anchor and low-anchor groups was $123. The
anchoring effect was above 30%, indicating that increasing the initial request by $100
broughtareturnof$30inaveragewillingnesstopay.
Similarorevenlargeranchoringeffectshavebeenobtainedinnumerousstudiesof
estimatesandofwillingnesstopay.Forexample,Frenchresidentsoftheheavilypolluted
Marseillesregionwereaskedwhatincreaseinlivingcoststheywouldacceptiftheycould
liveinalesspollutedregion.Theanchoringeffectwasover50%inthatstudy.Anchoring
effects are easily observed in online trading, where the same item is often offered at
different “buy now” prices. The “estimate” in fine-art auctions is also an anchor that
influencesthefirstbid.
There are situations in which anchoring appears reasonable. After all, it is not
surprisingthatpeoplewhoareaskeddifficultquestionsclutchatstraws,andtheanchoris
aplausiblestraw.IfyouknownexttonothingaboutthetreesofCaliforniaandareasked
whetheraredwoodcanbetallerthan1,200feet,youmightinferthatthisnumberisnot
toofarfromthetruth.Somebodywhoknowsthetrueheightthoughtupthatquestion,so
theanchormaybeavaluablehint.However,akeyfindingofanchoringresearchisthat
anchors that are obviously random can be just as effective as potentially informative
anchors. When we used a wheel of fortune to anchor estimates of the proportion of
AfricannationsintheUN,theanchoringindexwas44%,wellwithintherangeofeffects
observedwithanchorsthatcouldplausiblybetakenashints.Anchoringeffectsofsimilar
sizehavebeenobservedinexperimentsinwhichthelastfewdigitsoftherespondent’s
Social Security number was used as the anchor (e.g., for estimating the number of
physiciansintheircity).Theconclusionisclear:anchorsdonothavetheireffectsbecause
peoplebelievetheyareinformative.
The power of random anchors has been demonstrated in some unsettling ways.
Germanjudgeswithanaverageofmorethanfifteenyearsofexperienceonthebenchfirst
readadescriptionofawomanwhohadbeencaughtshoplifting,thenrolledapairofdice
thatwereloadedsoeveryrollresultedineithera3ora9.Assoonasthedicecametoa
stop,thejudgeswereaskedwhethertheywouldsentencethewomantoaterminprison
greaterorlesser,inmonths,thanthenumbershowingonthedice.Finally,thejudgeswere
instructed to specify the exact prison sentence they would give to the shoplifter. On
average,thosewhohadrolleda9saidtheywouldsentenceherto8months;thosewho
rolleda3saidthifAfricatheywouldsentenceherto5months;theanchoringeffectwas
50%.
UsesandAbusesofAnchors
By now you should be convinced that anchoring effects—sometimes due to priming,
sometimes to insufficient adjustment—are everywhere. The psychological mechanisms
thatproduceanchoringmakeusfarmoresuggestiblethanmostofuswouldwanttobe.
And of course there are quite a few people who are willing and able to exploit our
gullibility.
Anchoring effects explain why, for example, arbitrary rationing is an effective
marketingploy.Afewyearsago,supermarketshoppersinSiouxCity,Iowa,encountereda
salespromotionforCampbell’ssoupatabout10%offtheregularprice.Onsomedays,a
sign on the shelf said limit of 12 per person. On other days, the sign said no limit per
person.Shopperspurchasedan average of7canswhen thelimitwasinforce, twice as
manyastheyboughtwhenthelimitwasremoved.Anchoringisnotthesoleexplanation.
Rationingalsoimpliesthatthegoodsareflyingofftheshelves,andshoppersshouldfeel
some urgency about stocking up. But we also know that the mention of 12 cans as a
possible purchase would produce anchoring even if the number were produced by a
roulettewheel.
Weseethesamestrategyatworkinthenegotiationoverthepriceofahome,when
thesellermakesthefirstmovebysettingthelistprice.Asinmanyothergames,moving
first is an advantage in single-issue negotiations—for example, when price is the only
issue to be settled between a buyer and a seller. As you may have experienced when
negotiating for the first time in a bazaar, the initial anchor has a powerful effect. My
advicetostudentswhenItaughtnegotiationswasthatifyouthinktheothersidehasmade
an outrageous proposal, you should not come back with an equally outrageous
counteroffer,creatingagapthatwillbedifficulttobridgeinfurthernegotiations.Instead
youshouldmakeascene,stormoutorthreatentodoso,andmakeitclear—toyourselfas
wellastotheotherside—thatyouwillnotcontinuethenegotiationwiththatnumberon
thetable.
The psychologists Adam Galinsky and Thomas Mussweiler proposed more subtle
ways to resist the anchoring effect in negotiations. They instructed negotiators to focus
theirattentionandsearchtheirmemoryforargumentsagainsttheanchor.Theinstruction
to activate System 2 was successful. For example, the anchoring effect is reduced or
eliminated when the second mover focuses his attention on the minimal offer that the
opponentwouldaccept,oronthecoststotheopponentoffailingtoreachanagreement.
In general, a strategy of deliberately “thinking the opposite” may be a good defense
against anchoring effects, because it negates the biased recruitment of thoughts that
producestheseeffects.
Finally,tryyourhandatworkingouttheeffectofanchoringonaproblemofpublic
policy: the size of damages in personal injury cases. These awards are sometimes very
large.Businessesthatarefrequenttargetsofsuchlawsuits,suchashospitalsandchemical
companies, have lobbied to set a cap on the awards. Before you read this chapter you
might have thought that capping awards is certainly good for potential defendants, but
nowyoushouldnotbesosure.Considertheeffectofcappingawardsat$1million.This
rulewouldeliminatealllargerawards,buttheanchorwouldalsopullupthesizeofmany
awardsthatwouldotherwise be muchsmaller.It would almostcertainlybenefitserious
offendersandlargefirmsmuchmorethansmallones.
AnchoringandtheTwoSystems
TheeffectsofrandomanchorshavemuchtotellusabouttherelationshipbetweenSystem
1 and System 2. Anchoring effects have always been studied in tasks of judgment and
choicethatareultimatelycompletedbySystem2.However,System2worksondatathat
isretrievedfrommemory,inanautomaticandinvoluntaryoperationofSystem1.System
2isthereforesusceptibletothebiasinginfluenceofanchorsthatmakesomeinformation
easiertoretrieve.Furthermore,System2hasnocontrolovertheeffectandnoknowledge
of it. The participants who have been exposed to random or absurd anchors (such as
Gandhi’sdeathatage144)confidentlydenythatthisobviouslyuselessinformationcould
haveinfluencedtheirestimate,andtheyarewrong.
We saw in the discussion of the law of small numbers that a message, unless it is
immediately rejected as a lie, will have the same effect on the associative system
regardless of its reliability. The gist of the message is the story, which is based on
whateverinformationisavailable,evenifthequantityoftheinformationisslightandits
qualityispoor:WYSIATI.Whenyoureadastoryabouttheheroicrescueofawounded
mountainclimber,itseffectonyourassociativememoryismuchthesameifitisanews
report or the synopsis of a film. Anchoring results from this associative activation.
Whether the story is true, or believable, matters little, if at all. The powerful effect of
random anchors is an extreme case of this phenomenon, because a random anchor
obviouslyprovidesnoinformationatall.
EarlierIdiscussedthebewilderingvarietyofprimingeffects,inwhichyourthoughts
andbehaviormaybeinfluencedbystimulitowhichyoupaynoattentionatall,andeven
bystimuliofwhichyouarecompletelyunaware.Themainmoralofprimingresearchis
thatourthoughtsandourbehaviorareinfluenced,muchmorethanweknoworwant,by
the environment of the moment. Many people find the priming results unbelievable,
because they do not correspond to subjective experience. Many others find the results
upsetting, because they threaten the subjective sense of agency and autonomy. If the
content of a screen saver on an irrelevant computer can affect your willingness to help
strangers without your being aware of it, how free are you? Anchoring effects are
threateninginasimilarway.Youarealwaysawareoftheanchorandevenpayattentionto
it,butyoudonotknowhowitguidesandconstrainsyourthinking,becauseyoucannot
imagine how you would have thought if the anchor had been different (or absent).
However,youshouldassumethatanynumberthatisonthetablehashadananchoring
effectonyou,andifthestakesarehighyoushouldmobilizeyourself(yourSystem2)to
combattheeffect.
SpeakingofAnchors
“The firm we want to acquire sent us their business plan, with the revenue they
expect.Weshouldn’tletthatnumberinfluenceourthinking.Setitaside.”
“Plans are best-case scenarios. Let’s avoid anchoring on plans when we forecast
actualoutcomes.Thinkingaboutwaystheplancouldgowrongisonewaytodoit.”
“Ouraiminthenegotiationistogetthemanchoredonthisnumber.”
&st
“The defendant’s lawyers put in a frivolous reference in which they mentioned a
ridiculouslylowamountofdamages,andtheygotthejudgeanchoredonit!”
P
TheScienceofAvailability
AmosandIhadourmostproductiveyearin1971–72,whichwespentinEugene,Oregon.
WeweretheguestsoftheOregonResearchInstitute,whichhousedseveralfuturestarsof
allthefieldsinwhichweworked—judgment,decisionmaking,andintuitiveprediction.
Our main host was Paul Slovic, who had been Amos’s classmate at Ann Arbor and
remained a lifelong friend. Paul was on his way to becoming the leading psychologist
amongscholarsofrisk,apositionhehasheldfordecades,collectingmanyhonorsalong
theway.Paulandhiswife,Roz,introducedustolifeinEugene,andsoonweweredoing
whatpeopleinEugenedo—jogging,barbecuing,andtakingchildrentobasketballgames.
We also worked very hard, running dozens of experiments and writing our articles on
judgmentheuristics.AtnightIwroteAttentionandEffort.Itwasabusyyear.
One of our projects was the study of what we called the availability heuristic. We
thoughtofthatheuristicwhenweaskedourselveswhatpeopleactuallydowhentheywish
toestimatethefrequencyofacategory,suchas“peoplewhodivorceaftertheageof60”
or “dangerous plants.” The answer was straightforward: instances of the class will be
retrievedfrommemory,andifretrievaliseasyandfluent,thecategorywillbejudgedto
belarge.Wedefinedtheavailabilityheuristicastheprocessofjudgingfrequencyby“the
ease with which instances come to mind.” The statement seemed clear when we
formulatedit,buttheconceptofavailabilityhasbeenrefinedsincethen.Thetwo-system
approachhadnotyetbeendevelopedwhenwestudiedavailability,andwedidnotattempt
to determine whether this heuristic is a deliberate problem-solving strategy or an
automaticoperation.Wenowknowthatbothsystemsareinvolved.
Aquestionweconsideredearlywashowmanyinstancesmustberetrievedtogetan
impressionoftheeasewithwhichtheycometomind.Wenowknowtheanswer:none.
Foranexample,thinkofthenumberofwordsthatcanbeconstructedfromthetwosetsof
lettersbelow.
XUZONLCJM
TAPCERHOB
You knew almost immediately, without generating any instances, that one set offers far
morepossibilitiesthantheother,probablybyafactorof10ormore.Similarly,youdonot
needtoretrievespecificnewsstoriestohaveagoodideaoftherelativefrequencywith
whichdifferentcountrieshaveappearedinthenewsduringthepastyear(Belgium,China,
France,Congo,Nicaragua,Romania…).
Theavailabilityheuristic,likeotherheuristicsofjudgment,substitutesonequestion
foranother:youwishtoestimatethesizeseostcdofacategoryorthefrequencyofan
event, but you report an impression of the ease with which instances come to mind.
Substitutionofquestionsinevitablyproducessystematicerrors.Youcandiscoverhowthe
heuristicleadstobiasesbyfollowingasimpleprocedure:listfactorsotherthanfrequency
thatmakeiteasytocomeupwithinstances.Eachfactorinyourlistwillbeapotential
sourceofbias.Herearesomeexamples:
A salient event that attracts your attention will be easily retrieved from memory.
Divorces among Hollywood celebrities and sex scandals among politicians attract
muchattention,andinstanceswillcomeeasilytomind.Youarethereforelikelyto
exaggeratethefrequencyofbothHollywooddivorcesandpoliticalsexscandals.
Adramaticeventtemporarilyincreasestheavailabilityofitscategory.Aplanecrash
thatattractsmediacoveragewilltemporarilyalteryourfeelingsaboutthesafetyof
flying.Accidentsareonyourmind,forawhile,afteryouseeacarburningattheside
oftheroad,andtheworldisforawhileamoredangerousplace.
Personalexperiences,pictures,andvividexamplesaremoreavailablethanincidents
thathappenedtoothers,ormerewords,orstatistics.Ajudicialerrorthataffectsyou
willundermineyourfaithinthejusticesystemmorethanasimilarincidentyouread
aboutinanewspaper.
Resistingthislargecollectionofpotentialavailabilitybiasesispossible,buttiresome.
You must make the effort to reconsider your impressions and intuitions by asking such
questions as, “Is our belief that theft s by teenagers are a major problem due to a few
recentinstancesinourneighborhood?”or“CoulditbethatIfeelnoneedtogetaflushot
because none of my acquaintances got the flu last year?” Maintaining one’s vigilance
againstbiasesisachore—butthechancetoavoidacostlymistakeissometimesworththe
effort.
One of the best-known studies of availability suggests that awareness of your own
biases can contribute to peace in marriages, and probably in other joint projects. In a
famousstudy,spouseswereasked,“Howlargewasyourpersonalcontributiontokeeping
theplacetidy,inpercentages?”They also answeredsimilarquestionsabout “taking out
thegarbage,”“initiatingsocialengagements,”etc.Wouldtheself-estimatedcontributions
addupto100%,ormore,orless?Asexpected,theself-assessedcontributionsaddedupto
morethan 100%.The explanation isa simpleavailabilitybias: both spouses remember
theirownindividualeffortsandcontributionsmuchmoreclearlythanthoseoftheother,
andthedifferenceinavailabilityleadstoadifferenceinjudgedfrequency.Thebiasisnot
necessarilyself-serving:spousesalsooverestimatedtheircontributiontocausingquarrels,
althoughtoasmallerextentthantheircontributionstomoredesirableoutcomes.Thesame
biascontributestothecommonobservationthatmanymembersofacollaborativeteam
feeltheyhavedonemorethantheirshareandalsofeelthattheothersarenotadequately
gratefulfortheirindividualcontributions.
Iam generally notoptimisticabout thepotentialfor personal controlof biases, but
this is an exception. The opportunity for successful debiasing exists because the
circumstancesinwhichissuesofcreditallocationcomeupareeasytoidentify,themore
sobecausetensionsoftenarisewhenseveralpeopleatoncefeelthattheireffortsarenot
adequatelyrecognized.Themereobservationthatthereisusuallymorethan100%credit
togoaroundissometimessufficienttodefusethesituation.Inanyeve#82ght=nt,itisa
good thing for every individual to remember. You will occasionally do more than your
share, but it is useful to know that you are likely to have that feeling even when each
memberoftheteamfeelsthesameway.
ThePsychologyofAvailability
Amajor advancein theunderstanding ofthe availabilityheuristic occurredin theearly
1990s, when a group of German psychologists led by Norbert Schwarz raised an
intriguing question: How will people’s impressions of the frequency of a category be
affected by a requirement to list a specified number of instances? Imagine yourself a
subjectinthatexperiment:
First,listsixinstancesinwhichyoubehavedassertively.
Next,evaluatehowassertiveyouare.
Imagine that you had been asked for twelve instances of assertive behavior (a number
mostpeoplefinddifficult).Wouldyourviewofyourownassertivenessbedifferent?
Schwarzandhiscolleaguesobservedthatthetaskoflistinginstancesmayenhance
thejudgmentsofthetraitbytwodifferentroutes:
thenumberofinstancesretrieved
theeasewithwhichtheycometomind
Therequesttolisttwelveinstancespitsthetwodeterminantsagainsteachother.Onthe
one hand, you have just retrieved an impressive number of cases in which you were
assertive. On the other hand, while the first three or four instances of your own
assertivenessprobablycameeasilytoyou,youalmostcertainlystruggledtocomeupwith
thelastfewtocompleteasetoftwelve;fluencywaslow. Which will count more—the
amountretrievedortheeaseandfluencyoftheretrieval?
Thecontestyieldeda clear-cutwinner:peoplewhohad justlistedtwelveinstances
rated themselves as less assertive than people who had listed only six. Furthermore,
participants who had been asked to list twelve cases in which they had not behaved
assertivelyendedupthinkingofthemselvesasquiteassertive!Ifyoucannoteasilycome
upwithinstancesofmeekbehavior,youarelikelytoconcludethatyouarenotmeekat
all.Self-ratingsweredominatedbytheeasewithwhichexampleshadcometomind.The
experienceoffluentretrievalofinstancestrumpedthenumberretrieved.
An even more direct demonstration of the role of fluency was offered by other
psychologists in the same group. All the participants in their experiment listed six
instances of assertive (or nonassertive) behavior, while maintaining a specified facial
expression.“Smilers”wereinstructedtocontractthezygomaticusmuscle,whichproduces
a light smile; “frowners” were required to furrow their brow. As you already know,
frowningnormallyaccompaniescognitivestrainandtheeffectissymmetric:whenpeople
areinstructedtofrownwhiledoingatask,theyactuallytryharderandexperiencegreater
cognitivestrain.Theresearchersanticipatedthatthefrownerswouldhavemoredifficulty
retrievingexamplesofassertivebehaviorandwouldthereforeratethemselvesasrelatively
lackinginassertiveness.Andsoitwas.
Psychologists enjoy experiments that yield paradoxical results, and they have appliserv
heightedSchwarz’sdiscoverywithgusto.Forexample,people:
believe that they use their bicycles less often after recalling many rather than few
instances
are less confident in a choice when they are asked to produce more arguments to
supportit
arelessconfidentthataneventwasavoidableafterlistingmorewaysitcouldhave
beenavoided
arelessimpressedbyacarafterlistingmanyofitsadvantages
A professor at UCLA found an ingenious way to exploit the availability bias. He
askeddifferentgroupsofstudentstolistwaystoimprovethecourse,andhevariedthe
required number of improvements. As expected, the students who listed more ways to
improvetheclassratedithigher!
Perhapsthemostinterestingfindingofthisparadoxicalresearchisthattheparadoxis
notalwaysfound:peoplesometimesgobycontentratherthanbyeaseofretrieval.The
proofthatyoutrulyunderstandapatternofbehavioristhatyouknowhowtoreverseit.
Schwarz and his colleagues took on this challenge of discovering the conditions under
whichthisreversalwouldtakeplace.
Theeasewithwhichinstancesofassertivenesscometothesubject’smindchanges
duringthetask.Thefirstfewinstancesareeasy,butretrievalsoonbecomesmuchharder.
Of course, the subject also expects fluency to drop gradually, but the drop of fluency
betweensixandtwelveinstancesappearstobesteeperthantheparticipantexpected.The
results suggest that the participants make an inference: if I am having so much more
troublethanexpectedcomingupwithinstancesofmyassertiveness,thenIcan’tbevery
assertive.Notethatthisinferencerestsonasurprise—fluencybeingworsethanexpected.
The availability heuristic that the subjects apply is better described as an “unexplained
unavailability”heuristic.
Schwarz and his colleagues reasoned that they could disrupt the heuristic by
providing the subjects with an explanation for the fluency of retrieval that they
experienced.Theytoldtheparticipantstheywouldhearbackgroundmusicwhilerecalling
instancesandthatthemusicwouldaffectperformanceinthememorytask.Somesubjects
weretoldthat the musicwouldhelp, others weretoldto expect diminishedfluency. As
predicted, participants whose experience of fluency was “explained” did not use it as a
heuristic;thesubjectswhoweretoldthatmusicwouldmakeretrievalmoredifficultrated
themselves as equally assertive when they retrieved twelve instances as when they
retrievedsix.Othercoverstorieshavebeenusedwiththesameresult:judgmentsareno
longerinfluencedbyeaseofretrievalwhentheexperienceoffluencyisgivenaspurious
explanationbythepresenceofcurvedorstraighttextboxes,bythebackgroundcolorof
thescreen,orbyotherirrelevantfactorsthattheexperimentersdreamedup.
AsIhavedescribedit,theprocessthatleadstojudgmentbyavailabilityappearsto
involve a complex chain of reasoning. The subjects have an experience of diminishing
fluency as they produce instances. They evidently have expectations about the rate at
whichfluencydecreases,and thoseexpectationsarewrong: the difficultyofcomingup
with new instances increases more rapidly than they expect. It is the unexpectedly low
fluencythatcausespeoplewhowereaskedfortwelveinstancestodescribethemselvesas
unassertive. When the surprise is eliminated, low fluency no longer influences the
judgment.Theprocessappearstoconsistofasophisticatedriethesubjsetofinferences.Is
theautomaticSystem1capableofit?
Theansweristhatinfactnocomplexreasoningisneeded.Amongthebasicfeatures
ofSystem1isitsabilitytosetexpectationsandtobesurprisedwhentheseexpectations
areviolated.Thesystemalsoretrievespossiblecausesofasurprise,usuallybyfindinga
possiblecauseamongrecentsurprises.Furthermore,System2canresettheexpectations
ofSystem1onthefly,sothataneventthatwouldnormallybesurprisingisnowalmost
normal.Supposeyouaretoldthatthethree-year-oldboywholivesnextdoorfrequently
wearsatophatinhisstroller.Youwillbefarlesssurprisedwhenyouactuallyseehim
withhistophatthanyouwouldhavebeenwithoutthewarning.InSchwarz’sexperiment,
thebackgroundmusichasbeenmentionedasapossiblecauseofretrievalproblems.The
difficultyofretrievingtwelveinstancesisnolongerasurpriseandthereforeislesslikely
tobeevokedbythetaskofjudgingassertiveness.
Schwarzandhiscolleaguesdiscoveredthatpeoplewhoarepersonallyinvolvedinthe
judgmentaremorelikelytoconsiderthenumberofinstancestheyretrievefrommemory
andlesslikelytogobyfluency.Theyrecruitedtwogroupsofstudentsforastudyofrisks
to cardiac health. Half the students had a family history of cardiac disease and were
expectedtotakethetaskmoreseriouslythantheothers,whohadnosuchhistory.Allwere
askedtorecalleitherthreeoreightbehaviorsintheirroutinethatcouldaffecttheircardiac
health (some were asked for risky behaviors, others for protective behaviors). Students
with no family history of heart disease were casual about the task and followed the
availability heuristic. Students who found it difficult to find eight instances of risky
behaviorfeltthemselvesrelativelysafe,andthosewhostruggledtoretrieveexamplesof
safebehaviorsfeltthemselvesatrisk.Thestudentswithafamilyhistoryofheartdisease
showedtheoppositepattern—theyfeltsaferwhentheyretrievedmanyinstancesofsafe
behavior and felt greater danger when they retrieved many instances of risky behavior.
They were also more likely to feel that their future behavior would be affected by the
experienceofevaluatingtheirrisk.
The conclusion is that the ease with which instances come to mind is a System 1
heuristic, which is replaced by a focus on content when System 2 is more engaged.
Multiplelinesofevidenceconvergeontheconclusionthatpeoplewholetthemselvesbe
guidedbySystem1aremorestronglysusceptibletoavailabilitybiasesthanotherswho
areinastateofhighervigilance.Thefollowingaresomeconditionsinwhichpeople“go
withtheflow”andareaffectedmorestronglybyeaseofretrievalthanbythecontentthey
retrieved:
whentheyareengagedinanothereffortfultaskatthesametime
whentheyareinagoodmoodbecausetheyjustthoughtofahappyepisodeintheir
life
iftheyscorelowonadepressionscale
iftheyareknowledgeablenovicesonthetopicofthetask,incontrasttotrueexperts
whentheyscorehighonascaleoffaithinintuition
iftheyare(oraremadetofeel)powerful
Ifindthelastfindingparticularlyintriguing.Theauthorsintroducetheirarticlewitha
famousquote:“Idon’tspendalotoftimetakingpollsaroundtheworldtotellmewhatI
thinkistherightwaytoact.I’vejustgottoknowhowIfeel”(GeorgeeethewW.Bush,
November 2002). They go on to show that reliance on intuition is only in part a
personalitytrait.Merelyremindingpeopleofatimewhentheyhadpowerincreasestheir
apparenttrustintheirownintuition.
SpeakingofAvailability
“Becauseofthecoincidenceoftwoplanescrashinglastmonth,shenowprefersto
takethetrain.That’ssilly.Theriskhasn’treallychanged;itisanavailabilitybias.”
“Heunderestimatestherisksofindoorpollutionbecausetherearefewmediastories
onthem.That’sanavailabilityeffect.Heshouldlookatthestatistics.”
“Shehasbeenwatchingtoomanyspymoviesrecently,soshe’sseeingconspiracies
everywhere.”
“TheCEOhashadseveralsuccessesinarow,sofailuredoesn’tcomeeasilytoher
mind.Theavailabilitybiasismakingheroverconfident.”
P
Availability,Emotion,andRisk
Students of risk were quick to see that the idea of availability was relevant to their
concerns.Evenbeforeourworkwaspublished,theeconomistHowardKunreuther,who
was then in the early stages of a career that he has devoted to the study of risk and
insurance,noticedthatavailabilityeffectshelpexplainthepatternofinsurancepurchase
andprotectiveactionafterdisasters.Victimsandnearvictimsareveryconcernedaftera
disaster. After each significant earthquake, Californians are for a while diligent in
purchasinginsuranceandadoptingmeasuresofprotectionandmitigation.Theytiedown
their boiler to reduce quake damage, seal their basement doors against floods, and
maintainemergency supplies in goodorder.However, the memories of the disaster dim
over time, and so do worry and diligence. The dynamics of memory help explain the
recurrent cycles of disaster, concern, and growing complacency that are familiar to
studentsoflarge-scaleemergencies.
Kunreuther also observed that protective actions, whether by individuals or
governments, are usually designed to be adequate to the worst disaster actually
experienced.AslongagoaspharaonicEgypt,societieshavetrackedthehigh-watermark
of rivers that periodically flood—and have always prepared accordingly, apparently
assumingthatfloodswillnotrisehigherthantheexistinghigh-watermark.Imagesofa
worsedisasterdonotcomeeasilytomind.
AvailabilityandAffect
The most influential studies of availability biases were carried out by our friends in
Eugene,wherePaulSlovicandhislongtimecollaboratorSarahLichtensteinwerejoined
by our former student Baruch Fischhoff. They carried out groundbreaking research on
publicperceptionsofrisks,includingasurveythathasbecomethestandardexampleofan
availability bias. They asked participants in their survey to siIs th t#consider pairs of
causesofdeath:diabetesandasthma,orstrokeandaccidents.Foreachpair,thesubjects
indicated the more frequent cause and estimated the ratio of the two frequencies. The
judgments were compared to health statistics of the time. Here’s a sample of their
findings:
Strokes cause almost twice as many deaths as all accidents combined, but 80% of
respondentsjudgedaccidentaldeathtobemorelikely.
Tornadoeswereseenasmorefrequentkillersthanasthma,althoughthelattercause
20timesmoredeaths.
Deathbylightningwasjudgedlesslikelythandeathfrombotulismeventhoughitis
52timesmorefrequent.
Deathbydiseaseis18timesaslikelyasaccidentaldeath,butthetwowerejudged
aboutequallylikely.
Deathbyaccidentswasjudgedtobemorethan300timesmorelikelythandeathby
diabetes,butthetrueratiois1:4.
The lesson is clear: estimates of causes of death are warped by media coverage. The
coverageisitselfbiasedtowardnoveltyandpoignancy.Themediadonotjustshapewhat
the public is interested in, but also are shaped by it. Editors cannot ignore the public’s
demands that certain topics and viewpoints receive extensive coverage. Unusual events
(suchasbotulism)attractdisproportionateattentionandareconsequentlyperceivedasless
unusualthantheyreallyare.Theworldinourheadsisnotaprecisereplicaofreality;our
expectationsaboutthefrequencyofeventsaredistortedbytheprevalenceandemotional
intensityofthemessagestowhichweareexposed.
Theestimatesofcausesofdeathareanalmostdirectrepresentationoftheactivation
ofideasinassociativememory,andareagoodexampleofsubstitution.ButSlovicandhis
colleagueswereledtoadeeperinsight:theysawthattheeasewithwhichideasofvarious
risks come to mind and the emotional reactions to these risks are inextricably linked.
Frighteningthoughtsandimagesoccurtouswithparticularease,andthoughtsofdanger
thatarefluentandvividexacerbatefear.
Asmentionedearlier,Sloviceventuallydevelopedthenotionofanaffectheuristic,in
whichpeoplemakejudgmentsanddecisionsbyconsultingtheiremotions:DoIlikeit?Do
Ihateit?HowstronglydoIfeelaboutit?Inmanydomainsoflife,Slovicsaid,people
form opinions and make choices that directly express their feelings and their basic
tendencytoapproachoravoid,oftenwithoutknowingthattheyaredoingso.Theaffect
heuristicisaninstanceofsubstitution,inwhichtheanswertoaneasyquestion(HowdoI
feelaboutit?)servesasananswertoamuchharderquestion(WhatdoIthinkaboutit?).
Slovic and his colleagues related their views to the work of the neuroscientist Antonio
Damasio, who had proposed that people’s emotional evaluations of outcomes, and the
bodilystatesandtheapproachandavoidancetendenciesassociatedwiththem,allplaya
centralroleinguidingdecisionmaking.Damasioandhiscolleagueshaveobservedthat
people who do not display the appropriate emotions before they decide, sometimes
because of brain damage, also have an impaired ability to make good decisions. An
inabilitytobeguidedbya“healthyfear”ofbadconsequencesisadisastrousflaw.
In a compelling demonstration of the workings of the affect heuristic, Slovic’s
researchteamsurveyedopinionsaboutvarioustechnologies,includingwaterfluoridation,
chemicalplants,foodpreservatives,andcars,andaskedtheirrespondentstolistboththe
benefits>
The best part of the experiment came next. After completing the initial survey, the
respondents read brief passages with arguments in favor of various technologies. Some
were given arguments that focused on the numerous benefits of a technology; others,
arguments that stressed the low risks. These messages were effective in changing the
emotional appeal of the technologies. The striking finding was that people who had
receivedamessageextollingthebenefitsofatechnologyalsochangedtheirbeliefsabout
itsrisks.Althoughtheyhadreceivednorelevantevidence,thetechnologytheynowliked
morethanbeforewasalsoperceivedaslessrisky.Similarly,respondentswhoweretold
only that the risks of a technology were mild developed a more favorable view of its
benefits. The implication is clear: as the psychologist Jonathan Haidt said in another
context, “The emotional tail wags the rational dog.” The affect heuristic simplifies our
livesbycreatingaworldthatismuchtidierthanreality.Goodtechnologieshavefewcosts
intheimaginaryworldweinhabit,badtechnologieshavenobenefits,andalldecisionsare
easy. In the real world, of course, we often face painful tradeoffs between benefits and
costs.
ThePublicandtheExperts
PaulSlovicprobablyknowsmoreaboutthepeculiaritiesofhumanjudgmentofriskthan
any other individual. His work offers a picture of Mr. and Ms. Citizen that is far from
flattering:guidedbyemotionratherthanbyreason,easilyswayedbytrivialdetails,and
inadequatelysensitivetodifferencesbetweenlowandnegligiblylowprobabilities.Slovic
hasalsostudiedexperts,whoareclearlysuperiorindealingwithnumbersandamounts.
Expertsshowmanyofthesamebiasesastherestofusinattenuatedform,butoftentheir
judgmentsandpreferencesaboutrisksdivergefromthoseofotherpeople.
Differences between experts and the public are explained in part by biases in lay
judgments, but Slovic draws attention to situations in which the differences reflect a
genuineconflictofvalues.Hepointsoutthatexpertsoftenmeasurerisksbythenumberof
lives(or life-years)lost,while thepublic draws finer distinctions, forexample between
“gooddeaths”and“baddeaths,”orbetweenrandomaccidentalfatalitiesanddeathsthat
occurinthecourseofvoluntaryactivitiessuchasskiing.Theselegitimatedistinctionsare
oftenignoredinstatisticsthatmerelycountcases.Slovicarguesfromsuchobservations
that the public has a richer conception of risks than the experts do. Consequently, he
strongly resists the view that the experts should rule, and that their opinions shouldbe
accepted without question when they conflict with the opinions and wishes of other
citizens. When experts and the public disagree on their priorities, he says, “Each side
muiesstrespecttheinsightsandintelligenceoftheother.”
Inhisdesiretowrestsolecontrolofriskpolicyfromexperts,Slovichaschallenged
thefoundationoftheirexpertise:theideathatriskisobjective.
“Risk”doesnotexist“outthere,”independentofourmindsandculture,waitingtobe
measured. Human beings have invented the concept of “risk” to help them
understand and cope with the dangers and uncertainties of life. Although these
dangersarereal,thereisnosuchthingas“realrisk”or“objectiverisk.”
Toillustratehisclaim,Sloviclistsninewaysofdefiningthemortalityriskassociated
withthereleaseofatoxicmaterialintotheair,rangingfrom“deathpermillionpeople”to
“deathpermilliondollarsofproductproduced.”Hispointisthattheevaluationoftherisk
depends on the choice of a measure—with the obvious possibility that the choice may
havebeenguidedbyapreferenceforoneoutcomeoranother.Hegoesontoconcludethat
“definingriskisthusanexerciseinpower.”Youmightnothaveguessedthatonecanget
to such thorny policy issues from experimental studies of the psychology of judgment!
However, policy is ultimately about people, what they want and what is best for them.
Everypolicyquestioninvolvesassumptionsabouthumannature,inparticularaboutthe
choicesthatpeoplemaymakeandtheconsequencesoftheirchoicesforthemselvesand
forsociety.
AnotherscholarandfriendwhomIgreatlyadmire,CassSunstein,disagreessharply
withSlovic’sstanceonthedifferentviewsofexpertsandcitizens,anddefendstheroleof
experts as a bulwark against “populist” excesses. Sunstein is one of the foremost legal
scholarsintheUnitedStates,andshareswithotherleadersofhisprofessiontheattribute
ofintellectualfearlessness.Heknowshecanmasteranybodyofknowledgequicklyand
thoroughly, and he has mastered many, including both the psychology of judgment and
choice and issues of regulation and risk policy. His view is that the existing system of
regulation in the United States displays a very poor setting of priorities, which reflects
reaction to public pressures more than careful objective analysis. He starts from the
positionthatriskregulationandgovernmentinterventiontoreducerisksshouldbeguided
byrationalweightingofcostsandbenefits,andthatthenaturalunitsforthisanalysisare
thenumberoflivessaved(orperhapsthenumberoflife-yearssaved,whichgivesmore
weight to saving the young) and the dollar cost to the economy. Poor regulation is
wastefuloflivesandmoney,bothofwhichcanbemeasuredobjectively.Sunsteinhasnot
beenpersuadedbySlovic’sargumentthatriskanditsmeasurementissubjective.Many
aspectsofriskassessment are debatable,buthehas faith in theobjectivitythatmaybe
achievedbyscience,expertise,andcarefuldeliberation.
Sunstein came to believe that biased reactions to risks are an important source of
erraticandmisplacedprioritiesinpublicpolicy.Lawmakersandregulatorsmaybeoverly
responsivetotheirrationalconcernsofcitizens,bothbecauseofpoliticalsensitivityand
becausetheyarepronetothesamecognitivebiasesasothercitizens.
Sunstein and a collaborator, the jurist Timur Kuran, invented a name for the
mechanism through which biases flow into policy: the availability cascade. They
commentthatinthesocialcontext,“allheuristicsareequal,butavailabilityismoreequal
than the others.” They have in mind an expand Uned notion of the heuristic, in which
availability provides a heuristic for judgments other than frequency. In particular, the
importanceofanideaisoftenjudgedbythefluency(andemotionalcharge)withwhich
thatideacomestomind.
An availability cascade is a self-sustaining chain of events, which may start from
media reports of a relatively minor event and lead up to public panic and large-scale
governmentaction.Onsomeoccasions,amediastoryaboutariskcatchestheattentionof
a segment of the public, which becomes aroused and worried. This emotional reaction
becomes a story in itself, prompting additional coverage in the media, which in turn
produces greater concern and involvement. The cycle is sometimes sped along
deliberately by “availability entrepreneurs,” individuals or organizations who work to
ensureacontinuousflowofworryingnews.Thedangerisincreasinglyexaggeratedasthe
mediacompeteforattention-grabbingheadlines.Scientistsandotherswhotrytodampen
the increasing fear and revulsion attract little attention, most of it hostile: anyone who
claimsthatthedangerisoverstatedissuspectedofassociationwitha“heinouscover-up.”
The issue becomes politically important because it is on everyone’s mind, and the
response of the political system is guided by the intensity of public sentiment. The
availability cascade has now reset priorities. Other risks, and other ways that resources
couldbeappliedforthepublicgood,allhavefadedintothebackground.
Kuran and Sunstein focused on two examples that are still controversial: the Love
Canalaffairandtheso-calledAlarscare.InLoveCanal,buriedtoxicwastewasexposed
duringarainyseasonin1979,causingcontaminationofthewaterwellbeyondstandard
limits,aswellasafoulsmell.Theresidentsofthecommunitywereangryandfrightened,
andoneofthem,LoisGibbs,wasparticularlyactiveinanattempttosustaininterestinthe
problem.Theavailabilitycascade unfolded according tothe standard script.At its peak
thereweredailystoriesaboutLoveCanal,scientistsattemptingtoclaimthatthedangers
were overstated were ignored or shouted down, ABC News aired a program titled The
KillingGround,andemptybaby-sizecoffinswereparadedinfrontofthelegislature.A
largenumberofresidentswererelocatedatgovernmentexpense,andthecontroloftoxic
wastebecamethemajorenvironmentalissueofthe1980s.Thelegislationthatmandated
thecleanupoftoxicsites,calledCERCLA,establishedaSuperfundandisconsidereda
significant achievement of environmental legislation. It was also expensive, and some
haveclaimedthatthesameamountofmoneycouldhavesavedmanymorelivesifithad
beendirectedtootherpriorities.OpinionsaboutwhatactuallyhappenedatLoveCanalare
still sharply divided, and claims of actual damage to health appear not to have been
substantiated. Kuran and Sunstein wrote up the Love Canal story almost as a pseudo-
event, while on the other side of the debate, environmentalists still speak of the “Love
Canaldisaster.”
Opinions are also divided on the second example Kuran and Sunstein used to
illustratetheirconceptofanavailabilitycascade,theAlarincident,knowntodetractorsof
environmentalconcernsasthe“Alarscare”of1989.Alarisachemicalthatwassprayed
on apples to regulate their growth and improve their appearance. The scare began with
press stories that the chemical, when consumed in gigantic doses, caused cancerous
tumorsinratsandmice.Thestoriesunderstandablyfrightenedthepublic,andthosefears
encouraged more media coverage, the basic mechanism of an availability cascade. The
topicdominatedthenewsandproduceddramaticmediaeventssuchasthetestimonyof
theactressMerylStreepbeforeCongress.Theappleindustrysuofstainedlargelossesas
applesandappleproductsbecameobjectsoffear.KuranandSunsteinquoteacitizenwho
calledintoask“whetheritwassafertopourapplejuicedownthedrainortotakeittoa
toxic waste dump.” The manufacturer withdrew the product and the FDA banned it.
Subsequent research confirmed that the substance might pose a very small risk as a
possible carcinogen, but the Alar incident was certainly an enormous overreaction to a
minorproblem.Theneteffectoftheincidentonpublichealthwasprobablydetrimental
becausefewergoodappleswereconsumed.
TheAlartaleillustratesabasiclimitationintheabilityofourmindtodealwithsmall
risks: we either ignore them altogether or give them far too much weight—nothing in
between.Everyparentwhohasstayedupwaitingforateenagedaughterwhoislatefrom
apartywillrecognizethefeeling.Youmayknowthatthereisreally(almost)nothingto
worryabout,butyoucannothelpimagesofdisasterfromcomingtomind.AsSlovichas
argued,theamountofconcernisnotadequatelysensitivetotheprobabilityofharm;you
are imagining the numerator—the tragic story you saw on the news—and not thinking
aboutthedenominator. Sunsteinhascoined the phrase“probabilityneglect” todescribe
the pattern. The combination of probability neglect with the social mechanisms of
availability cascades inevitably leads to gross exaggeration of minor threats, sometimes
withimportantconsequences.
Intoday’sworld,terroristsarethemostsignificantpractitionersoftheartofinducing
availability cascades. With a few horrible exceptions such as 9/11, the number of
casualties from terror attacks is very small relative to other causes of death. Even in
countriesthathavebeentargetsofintensiveterrorcampaigns,suchasIsrael,theweekly
number of casualties almost never came close to the number of traffic deaths. The
differenceisin theavailabilityof thetworisks, theease and thefrequency with which
theycometomind.Gruesomeimages,endlesslyrepeatedinthemedia,causeeveryoneto
be on edge. As I know from experience, it is difficult to reason oneself into a state of
completecalm.TerrorismspeaksdirectlytoSystem1.
WheredoIcomedowninthedebatebetweenmyfriends?Availabilitycascadesare
real and they undoubtedly distort priorities in the allocation of public resources. Cass
Sunstein would seek mechanisms that insulate decision makers from public pressures,
lettingthe allocationofresourcesbedeterminedbyimpartialexpertswhohaveabroad
view of all risks and of the resources available to reduce them. Paul Slovic trusts the
expertsmuch lessandthepublicsomewhatmorethanSunsteindoes,andhe pointsout
that insulating the experts from the emotions of the public produces policies that the
publicwillreject—animpossiblesituationinademocracy.Bothareeminentlysensible,
andIagreewithboth.
I share Sunstein’s discomfort with the influence of irrational fears and availability
cascadesonpublicpolicyinthedomainofrisk.However,IalsoshareSlovic’sbeliefthat
widespreadfears,eveniftheyareunreasonable,shouldnotbeignoredbypolicymakers.
Rational or not, fear is painful and debilitating, and policy makers must endeavor to
protectthepublicfromfear,notonlyfromrealdangers.
Slovicrightlystressestheresistanceofthepublictotheideaofdecisionsbeingmade
by unelected and unaccountable experts. Furthermore, availability cascades may have a
long-termbenefitbycallingattentiontoclassesofrisksandbyincreasingtheoverallsize
of the risk-reduction budget. The Love Canal incident may have caused excessive
resources to be allocated to the management of toxic betwaste, but it also had a more
general effect in raising the priority level of environmental concerns. Democracy is
inevitablymessy,inpartbecausetheavailabilityandaffectheuristicsthatguidecitizens’
beliefs and attitudes are inevitably biased, even if they generally point in the right
direction.Psychologyshouldinformthedesignofriskpoliciesthatcombinetheexperts’
knowledgewiththepublic’semotionsandintuitions.
SpeakingofAvailabilityCascades
“She’sravingaboutaninnovationthathaslargebenefitsandnocosts.Isuspectthe
affectheuristic.”
“This is an availability cascade: a nonevent that is inflated by the media and the
publicuntilitfillsourTVscreensandbecomesallanyoneistalkingabout.”
P
TomW’sSpecialty
Havealookatasimplepuzzle:
Tom W is a graduate student at the main university in your state. Please rank the
followingninefieldsofgraduatespecializationinorderofthelikelihoodthatTomW
is now a student in each of these fields. Use 1 for the most likely, 9 for the least
likely.
businessadministration
computerscience
engineering
humanitiesandeducation
law
medicine
libraryscience
physicalandlifesciences
socialscienceandsocialwork
Thisquestioniseasy,andyouknewimmediatelythattherelativesizeofenrollment
inthedifferentfieldsisthekeytoasolution.Sofarasyouknow,TomWwaspickedat
randomfromthegraduatestudentsattheuniversity,likeasinglemarbledrawnfroman
urn.Todecidewhetheramarbleismorelikelytoberedorgreen,youneedtoknowhow
manymarblesofeachcolorthereareintheurn.Theproportionofmarblesofaparticular
kind is called a base rate. Similarly, the base rate of humanities and education in this
problemistheproportionofstudentsofthatfieldamongallthegraduatestudents.Inthe
absenceofspecificinformationaboutTomW,youwillgobythebaseratesandguessthat
heismorelikelytobeenrolledinhumanitiesandeducationthanincomputerscienceor
libraryscience,becausetherearemorestudentsoverallinthehumanitiesandeducation
than in the other two fields. Using base-rate information is the obvious move when no
otherinformationisprovided.
Nextcomesataskthathasnothingtodowithbaserates.
ThefollowingisapersonalitysketchofTomWwrittenduringTom’ssenioryearin
high school by a psychologist, on the basis of psychological tests of uncertain
validity:
TomWisofhighintelligence,althoughlackingintruecreativity.Hehasaneedfor
order and clarity, and for neat and tidy systems in which every detail finds its
appropriateplace.Hiswritingisratherdullandmechanical,occasionallyenlivened
bysomewhatcornypunsandflashesofimaginationofthesci-fitype.Hehasastrong
drive for competence. He seems to have little feel and little sympathy for other
people,anddoesnotenjoyinteractingwithothers.Self-centered,henonethelesshas
adeepmoralsense.
Now please take a sheet of paper and rank the nine fields of specialization listed
belowbyhowsimilarthedescriptionofTomWistothetypicalgraduatestudentin
eachofthefollowingfields.Use1forthemostlikelyand9fortheleastlikely.
You will get more out of the chapter if you give the task a quick try; reading the
report on Tom W is necessary to make your judgments about the various graduate
specialties.
This question too is straightforward. It requires you to retrieve, or perhaps to
construct,astereotypeofgraduatestudentsinthedifferentfields.Whentheexperiment
was first conducted, in the early 1970s, the average ordering was as follows. Yours is
probablynotverydifferent:
1. computerscience
2. engineering
3. businessadministration
4. physicalandlifesciences
5. libraryscience
6. law
7. medicine
8. humanitiesandeducation
9. socialscienceandsocialwork
You probably ranked computer science among the best fitting because of hints of
nerdiness (“corny puns”). In fact, the description of Tom W was written to fit that
stereotype.Anotherspecialtythatmostpeoplerankedhighisengineering(“neatandtidy
systems”).Youprobablythought thatTomWisnot a goodfitwith yourideaof social
scienceandsocialwork (“littlefeelandlittlesympathy for otherpeople”).Professional
stereotypes appear to have changed little in the nearly forty years since I designed the
descriptionofTomW.
Thetaskofrankingtheninecareersiscomplexandcertainlyrequiresthediscipline
andsequentialorganizationofwhichonlySystem2iscapable.However,thehintsplanted
inthedescription(cornypunsandothers)wereintendedtoactivateanassociationwitha
stereotype,anautomaticactivityofSystem1.
Theinstructions for thissimilarity taskrequired a comparisonof thedescription of
TomWtothestereotypesofthevariousfieldsofspecialization.Forthepurposesoftv>
IfyouexamineTomWagain,youwillseethatheisagoodfittostereotypesofsome
smallgroupsofstudents(computerscientists,librarians,engineers)andamuchpoorerfit
tothelargestgroups(humanitiesandeducation,socialscienceandsocialwork).Indeed,
the participants almost always ranked the two largest fields very low. Tom W was
intentionallydesignedasan“anti-base-rate”character,agoodfittosmallfieldsandapoor
fittothemostpopulatedspecialties.
PredictingbyRepresentativeness
Thethirdtaskinthesequencewasadministeredtograduatestudentsinpsychology,andit
isthecriticalone:rankthefieldsofspecializationinorderofthelikelihoodthatTomWis
nowagraduatestudentineachofthesefields.Themembersofthispredictiongroupknew
therelevantstatisticalfacts:theywerefamiliarwiththebaseratesofthedifferentfields,
and they knew that the source of Tom W’s description was not highly trustworthy.
However,weexpectedthemtofocusexclusivelyonthesimilarityofthedescriptiontothe
stereotypes—wecalleditrepresentativeness—ignoringboththebaseratesandthedoubts
abouttheveracityofthedescription.Theywouldthenrankthesmallspecialty—computer
science—as highly probable, because that outcome gets the highest representativeness
score.
AmosandIworkedhardduringtheyearwespentinEugene,andIsometimesstayed
in the office through the night. One of my tasks for such a night was to make up a
description that would pit representativeness and base rates against eachother. Tom W
wastheresultofmyefforts,andIcompletedthedescriptionintheearlymorninghours.
ThefirstpersonwhoshoweduptoworkthatmorningwasourcolleagueandfriendRobyn
Dawes, who was both a sophisticated statistician and a skeptic about the validity of
intuitivejudgment.Ifanyonewouldseetherelevanceofthebaserate,itwouldhavetobe
Robyn. I called Robyn over, gave him the question I had just typed, and asked him to
guessTomW’sprofession.Istillrememberhisslysmileashesaidtentatively,“computer
scientist?” That was a happy moment—even the mighty had fallen. Of course, Robyn
immediatelyrecognizedhismistake as soonasI mentioned“baserate,”but hehadnot
spontaneouslythoughtofit.Althoughheknewasmuchasanyoneabouttheroleofbase
rates in prediction, he neglected them when presented with the description of an
individual’spersonality.Asexpected,hesubstitutedajudgmentofrepresentativenessfor
theprobabilityhewasaskedtoassess.
AmosandIthencollectedanswerstothesamequestionfrom114graduatestudents
in psychology at three major universities, all of whom had taken several courses in
statistics.Theydidnotdisappointus.Theirrankingsoftheninefieldsbyprobabilitydid
notdifferfromratingsbysimilaritytothestereotype.Substitutionwasperfectinthiscase:
therewasnoindicationthattheparticipantsdidanythingelsebutjudgerepresentativeness.
Thequestionaboutprobability(likelihood)wasdifficult,butthequestionaboutsimilarity
waseasier,anditwasansweredinstead.Thisisaseriousmistake,becausejudgmentsof
similarityand probak tbilityare notconstrained by thesame logicalrules. Itis entirely
acceptable for judgments of similarity to be unaffected by base rates and also by the
possibilitythatthedescriptionwasinaccurate,butanyonewhoignoresbaseratesandthe
qualityofevidenceinprobabilityassessmentswillcertainlymakemistakes.
Theconcept“theprobabilitythatTomWstudiescomputerscience”isnotasimple
one.Logiciansandstatisticiansdisagreeaboutitsmeaning,andsomewouldsayithasno
meaningatall.Formanyexpertsitisameasureofsubjectivedegreeofbelief.Thereare
someeventsyouaresureof,forexample,thatthesunrosethismorning,andothersyou
considerimpossible,suchasthePacificOceanfreezingallatonce.Thentherearemany
events,suchasyournext-doorneighborbeingacomputerscientist,towhichyouassign
anintermediatedegreeofbelief—whichisyourprobabilityofthatevent.
Logicians and statisticians have developed competing definitions of probability, all
veryprecise. Forlaypeople, however,probability (asynonym of likelihood in everyday
language)isavaguenotion,relatedtouncertainty,propensity,plausibility,andsurprise.
Thevaguenessisnotparticulartothisconcept,norisitespeciallytroublesome.Weknow,
moreorless,whatwemeanwhenweuseawordsuchasdemocracyorbeauty andthe
peoplewearetalkingtounderstand,moreorless,whatweintendedtosay.Inalltheyears
Ispentaskingquestionsabouttheprobabilityofevents,nooneeverraisedahandtoask
me,“Sir,whatdoyoumeanbyprobability?”astheywouldhavedoneifIhadaskedthem
to assess a strange concept such as globability. Everyone acted as if they knew how to
answermyquestions,althoughweallunderstoodthatitwouldbeunfairtoaskthemforan
explanationofwhatthewordmeans.
Peoplewhoareaskedtoassessprobabilityarenotstumped,becausetheydonottryto
judge probability as statisticians and philosophers use the word. A question about
probabilityorlikelihoodactivatesamentalshotgun,evokinganswerstoeasierquestions.
One of the easy answers is an automatic assessment of representativeness—routine in
understandinglanguage.The(false)statementthat“ElvisPresley’sparentswantedhimto
beadentist”ismildlyfunnybecausethediscrepancybetweentheimagesofPresleyanda
dentistisdetectedautomatically.System1generatesanimpressionofsimilaritywithout
intendingtodoso.Therepresentativenessheuristicisinvolvedwhensomeonesays“She
willwintheelection;youcanseesheisawinner”or“Hewon’tgofarasanacademic;too
manytattoos.”Werelyonrepresentativenesswhenwejudgethepotentialleadershipofa
candidateforofficebytheshapeofhischinortheforcefulnessofhisspeeches.
Althoughitiscommon,predictionbyrepresentativenessisnotstatisticallyoptimal.
MichaelLewis’sbestsellingMoneyballisastoryabouttheinefficiencyofthismodeof
prediction. Professional baseball scouts traditionally forecast the success of possible
players in part by their build and look. The hero of Lewis’s book is Billy Beane, the
manageroftheOaklandA’s,whomadetheunpopulardecisiontooverrulehisscoutsand
to select players by the statistics of past performance. The players the A’s picked were
inexpensive, because other teams had rejected them for not looking the part. The team
soonachievedexcellentresultsatlowcost.
TheSinsofRepresentativeness
Judging probability byals representativeness has important virtues: the intuitive
impressions that it produces are often—indeed, usually—more accurate than chance
guesseswouldbe.
Onmostoccasions,peoplewhoactfriendlyareinfactfriendly.
Aprofessionalathletewhoisverytallandthinismuchmorelikelytoplaybasketball
thanfootball.
PeoplewithaPhDaremorelikelytosubscribetoTheNewYorkTimesthanpeople
whoendedtheireducationafterhighschool.
Youngmenaremorelikelythanelderlywomentodriveaggressively.
Inallthesecasesandinmanyothers,thereissometruthtothestereotypesthatgovern
judgments of representativeness, and predictions that follow this heuristic may be
accurate.Inothersituations,thestereotypesarefalseandtherepresentativenessheuristic
willmislead,especiallyifitcausespeopletoneglectbase-rateinformationthatpointsin
anotherdirection.Evenwhentheheuristic has somevalidity,exclusive reliance on itis
associatedwithgravesinsagainststatisticallogic.
Onesinofrepresentativenessisanexcessivewillingnesstopredicttheoccurrenceof
unlikely(low base-rate) events.Hereisanexample:youseeapersonreadingTheNew
YorkTimes on the New York subway. Which of the following is a better bet about the
readingstranger?
ShehasaPhD.
Shedoesnothaveacollegedegree.
RepresentativenesswouldtellyoutobetonthePhD,butthisisnotnecessarilywise.You
shouldseriouslyconsiderthe second alternative,becausemanymorenongraduates than
PhDs ride in New York subways. And if you must guess whether a woman who is
describedas“ashypoetrylover”studiesChineseliteratureorbusinessadministration,you
shouldoptforthelatteroption.EvenifeveryfemalestudentofChineseliteratureisshy
andlovespoetry,itisalmostcertainthattherearemorebashfulpoetryloversinthemuch
largerpopulationofbusinessstudents.
Peoplewithouttraininginstatisticsarequitecapableofusingbaseratesinpredictions
under some conditions. In the first version of the Tom W problem, which provides no
detailsabout him,it is obvious to everyonethat the probability of TomW’s being in a
particular field is simply the base rate frequency of enrollment in that field. However,
concernforbaseratesevidentlydisappearsassoonasTomW’spersonalityisdescribed.
Amos and I originally believed, on the basis of our early evidence, that base-rate
information will always be neglected when information about the specific instance is
available, but that conclusion was too strong. Psychologists have conducted many
experimentsinwhichbase-rateinformationisexplicitlyprovidedaspartoftheproblem,
andmanyoftheparticipantsareinfluencedbythosebaserates,althoughtheinformation
about the individual case is almost always weighted more than mere statistics. Norbert
Schwarz and his colleagues showed that instructing people to “think like a statistician”
enhancedtheuseofbase-rateinformation,whiletheinstructionto“thinklikeaclinician”
hadtheoppositeeffect.
AnexperimentthatwasconductedafewyearsagowithHarvardundergradutoates
yieldedafindingthatsurprisedme:enhancedactivationofSystem2causedasignificant
improvementofpredictiveaccuracyintheTomWproblem.Theexperimentcombinedthe
oldproblemwithamodernvariationofcognitivefluency.Halfthestudentsweretoldto
puffouttheircheeksduringthetask,whiletheothersweretoldtofrown.Frowning,aswe
haveseen,generallyincreasesthevigilanceofSystem2andreducesbothoverconfidence
andthe reliance onintuition.The studentswhopuffedouttheircheeks (anemotionally
neutral expression) replicated the original results: they relied exclusively on
representativenessandignoredthebaserates.Astheauthorshadpredicted,however,the
frownersdidshowsomesensitivitytothebaserates.Thisisaninstructivefinding.
When an incorrect intuitive judgment is made, System 1 and System 2 should both be
indicted. System 1 suggested the incorrect intuition, and System 2 endorsed it and
expressed it in a judgment. However, there are two possible reasons for the failure of
System 2—ignorance or laziness. Some people ignore base rates because they believe
them to be irrelevant in the presence of individual information. Others make the same
mistakebecausetheyarenotfocusedonthetask.Iffrowningmakesadifference,laziness
seems to be the proper explanation of base-rate neglect, at least among Harvard
undergrads.TheirSystem2“knows”thatbaseratesarerelevantevenwhentheyarenot
explicitlymentioned,butappliesthatknowledgeonlywhenitinvestsspecialeffortinthe
task.
Thesecondsinofrepresentativenessisinsensitivitytothequalityofevidence.Recall
theruleofSystem1:WYSIATI.IntheTomWexample,whatactivatesyourassociative
machineryisadescriptionofTom,whichmayormaynotbeanaccurateportrayal.The
statementthatTomW“haslittlefeelandlittlesympathyforpeople”wasprobablyenough
toconvinceyou(andmostotherreaders)thatheisveryunlikelytobeastudentofsocial
science or social work. But you were explicitly told that the description should not be
trusted!
Yousurelyunderstand in principlethatworthless informationshouldnot be treated
differentlyfromacompletelackofinformation,butWYSIATImakesitverydifficultto
applythatprinciple.Unlessyoudecideimmediatelytorejectevidence(forexample,by
determiningthatyoureceiveditfromaliar),yourSystem1willautomaticallyprocessthe
information available as if it were true. There is one thing you can do when you have
doubtsaboutthequalityoftheevidence:letyourjudgmentsofprobabilitystaycloseto
thebaserate.Don’texpectthisexerciseofdisciplinetobeeasy—itrequiresasignificant
effortofself-monitoringandself-control.
ThecorrectanswertotheTomWpuzzleisthatyoushouldstayveryclosetoyour
prior beliefs, slightly reducing the initially high probabilities of well-populated fields
(humanities and education; social science and social work) and slightly raising the low
probabilities of rare specialties (library science, computer science). You are not exactly
whereyouwouldbeifyouhadknownnothingatallaboutTomW,butthelittleevidence
youhaveisnottrustworthy,sothebaseratesshoulddominateyourestimates.
HowtoDisciplineIntuition
Your probability that it will rain tomorrow is your subjective degree of belief, but you
shouldnotletyourselfbelievewhatevercomestoyourmind.Tobeuseful,yourbeliefs
shouldbe constrainedby the logicof probability. Soif you believethat thereis a40%
chanceplethatitwillrainsometimetomorrow,youmustalsobelievethatthereisa60%
chanceitwillnotraintomorrow,andyoumustnotbelievethatthereisa50%chancethat
it will rain tomorrow morning. And if you believe that there is a 30% chance that
candidateXwillbeelectedpresident,andan80%chancethathewillbereelectedifhe
winsthefirsttime,thenyoumustbelievethatthechancesthathewillbeelectedtwiceina
roware24%.
Therelevant“rules”forcasessuchastheTomWproblemareprovidedbyBayesian
statistics.ThisinfluentialmodernapproachtostatisticsisnamedafteranEnglishminister
of the eighteenth century, the Reverend Thomas Bayes, who is credited with the first
majorcontributiontoalargeproblem:thelogicofhowpeopleshouldchangetheirmind
inthelightofevidence.Bayes’srulespecifieshowpriorbeliefs(intheexamplesofthis
chapter,baserates)shouldbecombinedwiththediagnosticityoftheevidence,thedegree
towhichitfavorsthehypothesisoverthealternative.Forexample,ifyoubelievethat3%
ofgraduatestudentsareenrolledincomputerscience(thebaserate),andyoualsobelieve
thatthedescriptionofTomWis4timesmorelikelyforagraduatestudentinthatfield
thaninotherfields,thenBayes’srulesaysyoumustbelievethattheprobabilitythatTom
Wisacomputerscientistisnow11%.Ifthebaseratehadbeen80%,thenewdegreeof
beliefwouldbe94.1%.Andsoon.
Themathematicaldetailsarenotrelevantinthisbook.Therearetwoideastokeepin
mindaboutBayesianreasoningandhowwetendtomessitup.Thefirstisthatbaserates
matter, even in the presence of evidence about the case at hand. This is often not
intuitively obvious. The second is that intuitive impressions of the diagnosticity of
evidenceareoftenexaggerated.ThecombinationofWYSIATIandassociativecoherence
tends to make us believe in the stories we spin for ourselves. The essential keys to
disciplinedBayesianreasoningcanbesimplysummarized:
Anchoryourjudgmentoftheprobabilityofanoutcomeonaplausiblebaserate.
Questionthediagnosticityofyourevidence.
Bothideasarestraightforward.ItcameasashocktomewhenIrealizedthatIwasnever
taughthowtoimplementthem,andthatevennowIfinditunnaturaltodoso.
SpeakingofRepresentativeness
“The lawn is well trimmed, the receptionist looks competent, and the furniture is
attractive,butthisdoesn’tmeanitisawell-managedcompany.Ihopetheboarddoes
notgobyrepresentativeness.”
“Thisstart-uplooksasifitcouldnotfail,butthebaserateofsuccessintheindustry
isextremelylow.Howdoweknowthiscaseisdifferent?”
“They keep making the same mistake: predicting rare events from weak evidence.
Whentheevidenceisweak,oneshouldstickwiththebaserates.”
“Iknowthisreportisabsolutelydamning,anditmaybebasedonsolidevidence,but
howsurearewe?Wemustallowforthatuncertaintyinourthinking.”
ht=“5%”>
P
Linda:LessIsMore
The best-known and most controversial of our experiments involved a fictitious lady
calledLinda.AmosandImadeuptheLindaproblemtoprovideconclusiveevidenceof
theroleofheuristicsinjudgmentandoftheirincompatibilitywithlogic.Thisishowwe
describedLinda:
Linda is thirty-one years old, single, outspoken, and very bright. She majored in
philosophy.Asastudent,shewasdeeplyconcernedwithissuesofdiscriminationand
socialjustice,andalsoparticipatedinantinucleardemonstrations.
The audiences who heard this description in the 1980s always laughed because they
immediatelyknewthatLindahadattendedtheUniversityofCaliforniaatBerkeley,which
was famous at the time for its radical, politically engaged students. In one of our
experimentswepresentedparticipantswithalistofeightpossiblescenariosforLinda.As
in the Tom W problem, some ranked the scenarios by representativeness, others by
probability.TheLindaproblemissimilar,butwithatwist.
Lindaisateacherinelementaryschool.
Lindaworksinabookstoreandtakesyogaclasses.
Lindaisactiveinthefeministmovement.
Lindaisapsychiatricsocialworker.
LindaisamemberoftheLeagueofWomenVoters.
Lindaisabankteller.
Lindaisaninsurancesalesperson.
Lindaisabanktellerandisactiveinthefeministmovement.
Theproblemshowsitsageinseveralways.TheLeagueofWomenVotersisnolongeras
prominentasitwas,andtheideaofafeminist“movement”soundsquaint,atestimonialto
thechangeinthestatusof women overthelastthirtyyears.Evenin the Facebookera,
however, it is still easy to guess the almost perfect consensus of judgments: Linda is a
very good fit for an active feminist, a fairly good fit for someone who works in a
bookstoreandtakesyoga classes—and averypoorfit for abanktelleror an insurance
salesperson.
Nowfocusonthecriticalitemsinthelist:DoesLindalookmorelikeabankteller,or
more like a bank teller who is active in the feminist movement? Everyone agrees that
Lindafitstheideaofa“feministbankteller”betterthanshefitsthestereotypeofbank
tellers.Thestereotypicalbanktellerisnotafeministactivist,andaddingthatdetailtothe
descriptionmakesforamorecoherentstory.
The twist comes in the judgments of likelihood, because there is a logical relation
between the two scenarios. Think in terms of Venn diagrams. The set of feminist bank
tellers is wholly included in the set of bank tellers, as every feminist bank teller
is0%“ustworaban0%”wabankteller.ThereforetheprobabilitythatLindaisafeminist
banktellermustbelowerthantheprobabilityofherbeingabankteller.Whenyouspecify
apossibleeventingreaterdetailyoucanonlyloweritsprobability.Theproblemtherefore
setsupaconflictbetweentheintuitionofrepresentativenessandthelogicofprobability.
Our initial experiment was between-subjects. Each participant saw a set of seven
outcomes that included only one of the critical items (“bank teller” or “feminist bank
teller”).Somerankedtheoutcomesbyresemblance,othersbylikelihood.Asinthecaseof
TomW,theaveragerankingsbyresemblanceandbylikelihoodwereidentical;“feminist
bankteller”rankedhigherthan“bankteller”inboth.
Thenwetooktheexperimentfurther,usingawithin-subjectdesign.Wemadeupthe
questionnaireasyousawit,with“bankteller”inthesixthpositioninthelistand“feminist
bankteller”asthelastitem.Wewereconvincedthatsubjectswouldnoticethe relation
betweenthetwooutcomes,andthattheirrankingswouldbeconsistentwithlogic.Indeed,
we were so certain of this that we did not think it worthwhile to conduct a special
experiment.My assistantwas runninganother experimentin thelab, andshe askedthe
subjectstocompletethenewLindaquestionnairewhilesigningout,justbeforetheygot
paid.
Abouttenquestionnaireshadaccumulatedinatrayonmyassistant’sdeskbeforeI
casuallyglancedatthemandfoundthatallthesubjectshadranked“feministbankteller”
as more probable than “bank teller.” I was so surprised that I still retain a “flashbulb
memory”ofthegraycolorofthemetaldeskandofwhereeveryonewaswhenImadethat
discovery.IquicklycalledAmosingreatexcitementtotellhimwhatwehadfound:we
hadpittedlogicagainstrepresentativeness,andrepresentativenesshadwon!
Inthelanguageofthisbook,wehadobservedafailureofSystem2:ourparticipants
hadafairopportunitytodetecttherelevanceofthelogicalrule,sincebothoutcomeswere
includedinthesameranking.Theydidnottakeadvantageofthatopportunity.Whenwe
extendedtheexperiment,wefoundthat89%oftheundergraduatesinoursampleviolated
the logic of probability. We were convinced that statistically sophisticated respondents
would do better, so we administered the same questionnaire to doctoral students in the
decision-scienceprogramoftheStanfordGraduateSchoolofBusiness,allofwhomhad
taken several advanced courses in probability, statistics, and decision theory. We were
surprised again: 85% of these respondents also ranked “feminist bank teller” as more
likelythan“bankteller.”
Inwhatwelaterdescribedas“increasinglydesperate”attemptstoeliminatetheerror,
weintroducedlargegroupsofpeopletoLindaandaskedthemthissimplequestion:
Whichalternativeismoreprobable?
Lindaisabankteller.
Lindaisabanktellerandisactiveinthefeministmovement.
Thisstarkversionof the problemmadeLindafamousin some circles,anditearned us
yearsofcontroversy.About85%to90%ofundergraduatesatseveralmajoruniversities
chose the second option, contrary to logic. Remarkably, the sinners seemed to have no
shame. When I asked my large undergraduatnite class in some indignation, “Do you
realize that you have violated an elementary logical rule?” someone in the back row
shouted,“Sowhat?”andagraduatestudentwhomadethesameerrorexplainedherselfby
saying,“Ithoughtyoujustaskedformyopinion.”
Thewordfallacyisused,ingeneral,whenpeoplefailtoapplyalogicalrulethatis
obviouslyrelevant.AmosandIintroducedtheideaofaconjunctionfallacy,whichpeople
commitwhentheyjudgeaconjunctionoftwoevents(here,banktellerandfeminist)tobe
moreprobablethanoneoftheevents(bankteller)inadirectcomparison.
AsintheMüller-Lyerillusion,thefallacyremainsattractiveevenwhenyourecognize
it for what it is. The naturalist Stephen Jay Gould described his own struggle with the
Linda problem. He knew the correct answer, of course, and yet, he wrote, “a little
homunculusinmyheadcontinuestojumpupanddown,shoutingatme—‘butshecan’t
justbe abank teller; readthe description.’”The little homunculusis ofcourse Gould’s
System 1 speaking to him in insistent tones. (The two-system terminology had not yet
beenintroducedwhenhewrote.)
The correct answer to the short version of the Linda problem was the majority
response in only one of our studies: 64% of a group of graduate students in the social
sciences at Stanford and at Berkeley correctly judged “feminist bank teller” to be less
probablethan“bankteller.”Intheoriginalversionwitheightoutcomes(shownabove),
only15%ofasimilargroupofgraduatestudentshadmadethatchoice.Thedifferenceis
instructive.Thelongerversionseparatedthetwocriticaloutcomesbyaninterveningitem
(insurance salesperson), and the readers judged each outcome independently, without
comparing them. The shorter version, in contrast, required an explicit comparison that
mobilizedSystem2andallowedmostofthestatisticallysophisticatedstudentstoavoid
the fallacy. Unfortunately, we did not explore the reasoning of the substantial minority
(36%)ofthisknowledgeablegroupwhochoseincorrectly.
Thejudgments ofprobability thatour respondentsoffered,inboth the TomWand
Lindaproblems,correspondedpreciselytojudgmentsofrepresentativeness(similarityto
stereotypes).Representativenessbelongstoaclusterofcloselyrelatedbasicassessments
thatarelikelytobegeneratedtogether.Themostrepresentativeoutcomescombinewith
thepersonalitydescriptiontoproducethemostcoherentstories.Themostcoherentstories
arenotnecessarilythemostprobable,buttheyareplausible,andthenotionsofcoherence,
plausibility,andprobabilityareeasilyconfusedbytheunwary.
The uncritical substitution of plausibility for probability has pernicious effects on
judgmentswhenscenariosareusedastoolsofforecasting.Considerthesetwoscenarios,
whichwerepresentedtodifferentgroups,witharequesttoevaluatetheirprobability:
AmassivefloodsomewhereinNorthAmericanextyear,inwhichmorethan1,000
peopledrown
AnearthquakeinCaliforniasometimenextyear,causingafloodinwhichmorethan
1,000peopledrown
The California earthquake scenario is more plausible than the North America scenario,
although its probability is certainly smaller. As expected, probability judgments were
higherfor thericher andmore entdetailedscenario, contraryto logic.This isa trapfor
forecastersandtheirclients:addingdetailtoscenariosmakesthemmorepersuasive,but
lesslikelytocometrue.
Toappreciatetheroleofplausibility,considerthefollowingquestions:
Whichalternativeismoreprobable?
Markhashair.
Markhasblondhair.
and
Whichalternativeismoreprobable?
Janeisateacher.
Janeisateacherandwalkstowork.
ThetwoquestionshavethesamelogicalstructureastheLindaproblem,buttheycauseno
fallacy,becausethemoredetailedoutcomeisonlymoredetailed—itisnotmoreplausible,
ormorecoherent,orabetterstory.Theevaluationofplausibilityandcoherencedoesnot
suggestandanswertotheprobabilityquestion.Intheabsenceofacompetingintuition,
logicprevails.
LessIsMore,SometimesEvenInJointEvaluation
ChristopherHsee,oftheUniversityofChicago,askedpeopletopricesetsofdinnerware
offeredinaclearancesaleinalocalstore,wheredinnerwareregularlyrunsbetween$30
and$60.Therewerethreegroupsinhisexperiment.Thedisplaybelowwasshowntoone
group;Hseelabelsthatjointevaluation,becauseitallowsacomparisonofthetwosets.
Theothertwogroupswereshownonlyoneofthetwosets;thisissingleevaluation.Joint
evaluationisawithin-subjectexperiment,andsingleevaluationisbetween-subjects.
SetA:40pieces SetB:24pieces
Dinnerplates 8,allingoodcondition 8,allingoodcondition
Soup/saladbowls 8,allingoodcondition 8,allingoodcondition
Dessertplates 8,allingoodcondition 8,allingoodcondition
Cups 8,2ofthembroken
Saucers 8,7ofthembroken
Assumingthatthedishesinthetwosetsareofequalquality,whichisworthmore?
Thisquestioniseasy.YoucanseethatSetAcontainsallthedishesofSetB,andseven
additional intact dishes, and it must be valued more. Indeed, the participants in Hsee’s
jointevaluationexperimentwerewillingtopayalittlemoreforSetAthanforSetB:$32
versus$30.
Theresultsreversedinsingleevaluation,whereSetBwaspricedmuchhigherthan
SetA:$33versus$23.Weknowwhythishappened.Sets(includingdinnerwaresets!)are
representedbynormsandprototypes.Youcansenseimmediatelythattheaveragevalueof
thedishesismuchlowerforSetAthanforSetB,becausenoonewantstopayforbroken
dishes.Iftheaverage dominates theevaluation,itisnot surprising thatSetBisvalued
more.Hseecalledtheresultingpatternlessismore.Byremoving16itemsfromSetA(7
ofthemintact),itsvalueisimproved.
Hsee’s finding was replicated by the experimental economist John List in a real
marketforbaseballcards.Heauctionedsetsoftenhigh-valuecards,andidenticalsetsto
which three cards of modest value were added. As in the dinnerware experiment, the
largersetswerevaluedmorethanthesmalleronesinjointevaluation,butlessinsingle
evaluation.Fromtheperspectiveofeconomictheory,thisresultistroubling:theeconomic
value of a dinnerware set or of a collection of baseball cards is a sum-like variable.
Addingapositivelyvalueditemtothesetcanonlyincreaseitsvalue.
The Linda problem and the dinnerware problem have exactly the same structure.
Probability,likeeconomicvalue,isasum-likevariable,asillustratedbythisexample:
probability(Lindaisateller)=probability(Lindaisfeministteller)+probability(Linda
isnon-feministteller)
Thisisalsowhy,asinHsee’sdinnerwarestudy,singleevaluationsoftheLindaproblem
produce a less-is-more pattern. System 1 averages instead of adding, so when the non-
feministbanktellersareremovedfromtheset,subjectiveprobabilityincreases.However,
thesum-likenature ofthevariable islessobvious forprobabilitythan formoney.As a
result, joint evaluation eliminates the error only in Hsee’s experiment, not in the Linda
experiment.
Linda was not the only conjunction error that survived joint evaluation. We found
similarviolationsoflogicinmanyotherjudgments.Participantsinoneofthesestudies
wereaskedtorankfourpossibleoutcomesofthenextWimbledontournamentfrommost
toleastprobable.BjörnBorgwasthedominanttennisplayerofthedaywhenthestudy
wasconducted.Theseweretheoutcomes:
A.Borgwillwinthematch.
B.Borgwilllosethefirstset.
C.Borgwilllosethefirstsetbutwinthematch.
D.Borgwillwinthefirstsetbutlosethematch.
ThecriticalitemsareBandC.Bisthemoreinclusiveeventanditsprobabilitymustbe
higherthanthatofaneventitincludes.Contrarytologic,butnottorepresentativenessor
plausibility,72%assignedBalowerprobabilitythanC—anotherinstanceoflessismore
in a direct comparison. Here si again, the scenario that was judged more probable was
unquestionablymoreplausible,amorecoherentfitwithallthatwasknownaboutthebest
tennisplayerintheworld.
To head off the possible objection that the conjunction fallacy is due to a
misinterpretation of probability, we constructed a problem that required probability
judgments,butinwhichtheeventswerenotdescribedinwords,andthetermprobability
didnot appear atall. We toldparticipantsabout a regularsix-sided die withfour green
faces and two red faces, which would be rolled 20 times. They were shown three
sequences of greens (G) and reds (R), and were asked to choose one. They would
(hypothetically)win$25iftheirchosensequenceshowedup.Thesequenceswere:
1. RGRRR
2. GRGRRR
3. GRRRRR
Because the die has twice as many green as red faces, the first sequence is quite
unrepresentative—likeLindabeingabankteller.Thesecondsequence,whichcontainssix
tosses,isabetterfittowhatwewouldexpectfromthisdie,becauseitincludestwoG’s.
However, this sequence was constructed by adding a G to the beginning of the first
sequence,soitcanonlybelesslikelythanthefirst.Thisisthenonverbalequivalentto
Lindabeingafeministbankteller.AsintheLindastudy,representativenessdominated.
Almosttwo-thirdsofrespondentspreferredtobetonsequence2ratherthanonsequence
1.Whenpresentedwithargumentsforthetwochoices,however,alargemajorityfound
thecorrectargument(favoringsequence1)moreconvincing.
Thenextproblemwasabreakthrough,becausewefinallyfoundaconditioninwhich
theincidenceoftheconjunctionfallacywasmuchreduced.Twogroupsofsubjectssaw
slightlydifferentvariantsofthesameproblem:
Theincidenceoferrorswas65%inthegroupthatsawtheproblemontheleft,andonly
25%inthegroupthatsawtheproblemontheright.
Why is the question “How many of the 100 participants…” so much easier than
“Whatpercentage…”?Alikelyexplanationisthatthereferenceto100individualsbrings
aspatialrepresentationtomind.Imaginethatalargenumberofpeopleareinstructedto
sortthemselvesintogroupsinaroom:“ThosewhosenamesbeginwiththelettersAtoL
are told to gather in the front left corner.” They are then instructed to sort themselves
further.Therelationofinclusionisnowobvious,andyoucanseethatindividualswhose
namebeginswithCwillbeasubsetofthecrowdinthefrontleftcorner.Inthemedical
surveyquestion,heartattackvictimsendupinacorneroftheroom,andsomeofthemare
less than 55 years old. Not everyone will share this particular vivid imagery, but many
subsequent experiments have shown that the frequency representation, as it is known,
makesiteasytoappreciatethatonegroupiswhollyincludedintheother.Thesolutionto
the puzzle appears to be that a question phrased as “how many?” makes you think of
individuals,butthesamequestionphrasedas“whatpercentage?”doesnot.
What have we learned from these studies about the workings of System 2? One
conclusion, which is not new, is that System 2 is not impressively alert. The
undergraduatesandgraduatestudentswhoparticipatedinourthastudiesoftheconjunction
fallacycertainly“knew”thelogicofVenndiagrams,buttheydidnotapplyitreliablyeven
whenalltherelevantinformationwaslaidoutinfrontofthem.Theabsurdityoftheless-
is-morepatternwasobviousinHsee’sdinnerwarestudyandwaseasilyrecognizedinthe
“howmany?”representation,butitwasnotapparenttothethousandsofpeoplewhohave
committedtheconjunctionfallacyintheoriginalLindaproblemandinotherslikeit.Inall
thesecases,theconjunctionappearedplausible,andthatsufficedforanendorsementof
System2.
ThelazinessofSystem2ispartofthestory.Iftheirnextvacationhaddependedonit,
andiftheyhadbeengivenindefinitetimeandtoldtofollowlogicandnottoansweruntil
theyweresureoftheiranswer,Ibelievethatmostofoursubjectswouldhaveavoidedthe
conjunction fallacy. However, their vacation did not depend on a correct answer; they
spentverylittletimeonit,andwerecontenttoanswerasiftheyhadonlybeen“askedfor
theiropinion.”ThelazinessofSystem2isanimportantfactoflife,andtheobservation
thatrepresentativenesscanblocktheapplicationofanobviouslogicalruleisalsoofsome
interest.
TheremarkableaspectoftheLindastoryisthecontrasttothebroken-dishesstudy.
Thetwoproblemshavethesamestructure,butyielddifferentresults.Peoplewhoseethe
dinnerware set that includes broken dishes put a very low price on it; their behavior
reflects a rule of intuition. Others who see both sets at once apply the logical rule that
more dishes can only add value. Intuition governs judgments in the between-subjects
condition;logicrulesinjointevaluation.IntheLindaproblem,incontrast,intuitionoften
overcamelogiceveninjointevaluation,althoughweidentifiedsomeconditionsinwhich
logicprevails.
AmosandIbelievedthattheblatantviolationsofthelogicofprobabilitythatwehad
observedintransparentproblemswereinterestingandworthreportingtoourcolleagues.
Wealsobelievedthattheresultsstrengthenedourargumentaboutthepowerofjudgment
heuristics, and that they would persuade doubters. And in this we were quite wrong.
Instead,theLindaproblembecameacasestudyinthenormsofcontroversy.
TheLindaproblemattractedagreatdealofattention,butitalsobecameamagnetfor
critics of our approach to judgment. As we had already done, researchers found
combinations of instructions and hints that reduced the incidence of the fallacy; some
arguedthat,inthecontextoftheLindaproblem,itisreasonableforsubjectstounderstand
the word “probability” as if it means “plausibility.” These arguments were sometimes
extended to suggest that our entire enterprise was misguided: if one salient cognitive
illusion could be weakened or explained away, others could be as well. This reasoning
neglects the unique feature of the conjunction fallacy as a case of conflict between
intuitionandlogic.Theevidencethatwehadbuiltupforheuristicsfrombetween-subjects
experiment(includingstudiesofLinda)wasnotchallenged—itwassimplynotaddressed,
anditssaliencewasdiminishedbytheexclusivefocusontheconjunctionfallacy.Thenet
effectoftheLindaproblemwas an increase in the visibilityofourworktothegeneral
public,andasmalldentinthecredibilityofourapproachamongscholarsinthefield.This
wasnotatallwhatwehadexpected.
Ifyouvisitacourtroomyouwillobservethatlawyersapplytwostylesofcriticism:to
demolishacasetheyraisedoubtsaboutthestrongestargumentsthatfavorit;todiscredita
witness,theyfocusontheweakestpartofthetestimony.Thefocusonweaknessesisalso
normalinpoliticaverldebates.Idonotbelieveitisappropriateinscientificcontroversies,
butIhavecometoacceptasafactoflifethatthenormsofdebateinthesocialsciencesdo
notprohibitthepoliticalstyleofargument,especiallywhenlargeissuesareatstake—and
theprevalenceofbiasinhumanjudgmentisalargeissue.
SomeyearsagoIhadafriendlyconversationwithRalphHertwig,apersistentcritic
of the Linda problem, with whom I had collaborated in a vain attempt to settle our
differences. I asked him why he and others had chosen to focus exclusively on the
conjunctionfallacy, ratherthan onother findingsthat provided strongersupport forour
position. He smiled as he answered, “It was more interesting,” adding that the Linda
problemhadattractedsomuchattentionthatwehadnoreasontocomplain.
SpeakingofLessisMore
“They constructed a very complicated scenario and insisted on calling it highly
probable.Itisnot—itisonlyaplausiblestory.”
“They added a cheap gift to the expensive product, and made the whole deal less
attractive.Lessismoreinthiscase.”
“Inmostsituations,adirectcomparisonmakespeoplemorecarefulandmorelogical.
Butnotalways.Sometimesintuitionbeatslogicevenwhenthecorrectanswerstares
youintheface.”
P
CausesTrumpStatistics
Considerthefollowingscenarioandnoteyourintuitiveanswertothequestion.
Acabwasinvolvedinahit-and-runaccidentatnight.
Twocabcompanies,theGreenandtheBlue,operateinthecity.
Youaregiventhefollowingdata:
85%ofthecabsinthecityareGreenand15%areBlue.
AwitnessidentifiedthecabasBlue.Thecourttested the reliability of thewitness
underthecircumstancesthatexistedonthenightoftheaccidentandconcludedthat
thewitnesscorrectlyidentifiedeachoneofthetwocolors80%ofthetimeandfailed
20%ofthetime.
What is the probability that the cab involved in the accident was Blue rather than
Green?
ThisisastandardproblemofBayesianinference.Therearetwoitemsofinformation:a
baserateandtheimperfectlyreliabletestimonyofawitness.Intheabsenceofawitness,
theprobabilityoftheguiltycabbeingBlueis15%,whichisthebaserateofthatoutcome.
Ifthetwocabcompanieshadbeenequallylarge,thebaseratewouldbeuninformativeand
youwouldconsideronlythereliabilityofthewitness,%”>ourw
CausalStereotypes
Nowconsideravariationofthesamestory,inwhichonlythepresentationofthebaserate
hasbeenaltered.
Youaregiventhefollowingdata:
Thetwocompaniesoperatethesamenumberofcabs,butGreencabsareinvolvedin
85%ofaccidents.
Theinformationaboutthewitnessisasinthepreviousversion.
The two versions of the problem are mathematically indistinguishable, but they are
psychologicallyquitedifferent.Peoplewhoreadthefirstversiondonotknowhowtouse
the base rate and often ignore it. In contrast, people who see the second version give
considerableweighttothebaserate,andtheiraveragejudgmentisnottoofarfromthe
Bayesiansolution.Why?
Inthefirstversion,thebaserateofBluecabsisastatisticalfactaboutthecabsinthe
city. A mind that is hungry for causal stories finds nothing to chew on: How does the
numberofGreenandBluecabsinthecitycausethiscabdrivertohitandrun?
Inthesecondversion,incontrast,thedriversofGreencabscausemorethan5times
asmanyaccidentsastheBluecabsdo.Theconclusionisimmediate:theGreendrivers
mustbe a collection of recklessmadmen! Youhave now formed a stereotypeof Green
recklessness, which you apply to unknown individual drivers in the company. The
stereotypeiseasilyfittedintoacausalstory,becauserecklessnessisacausallyrelevant
factaboutindividualcabdrivers.Inthisversion,therearetwocausalstoriesthatneedtobe
combinedorreconciled.Thefirstisthehitandrun,whichnaturallyevokestheideathata
reckless Green driver was responsible. The second is the witness’s testimony, which
stronglysuggeststhecabwasBlue.Theinferencesfromthetwostoriesaboutthecolorof
thecar arecontradictory andapproximately canceleach other.The chancesfor thetwo
colorsareaboutequal(theBayesianestimateis41%,reflectingthefactthatthebaserate
ofGreencabsisalittlemoreextremethanthereliabilityofthewitnesswhoreporteda
Bluecab).
The cab example illustrates two types of base rates. Statistical base rates are facts
about a population to which a case belongs, but they are not relevant to the individual
case.Causalbaserateschangeyourviewofhowtheindividualcasecametobe.Thetwo
typesofbase-rateinformationaretreateddifferently:
Statistical base rates are generally underweighted, and sometimes neglected
altogether,whenspecificinformationaboutthecaseathandisavailable.
Causalbaseratesaretreatedasinformationabouttheindividualcaseandareeasily
combinedwithothercase-specificinformation.
The causal version of the cab problem had the form of a stereotype: Green drivers are
dangerous. Stereotypes are statements about the group that are (at least tentatively)
acceptedasfactsabouteverymember.Helyrearetwoexamples:
Mostofthegraduatesofthisinner-cityschoolgotocollege.
InterestincyclingiswidespreadinFrance.
Thesestatementsarereadilyinterpretedassettingupapropensityinindividualmembers
ofthe group, andthey fitina causalstory. Manygraduates of thisparticular inner-city
schoolareeagerandabletogotocollege,presumablybecauseofsomebeneficialfeatures
oflifeinthatschool.ThereareforcesinFrenchcultureandsociallifethatcausemany
Frenchmentotakeaninterestincycling.Youwillberemindedofthesefactswhenyou
thinkaboutthelikelihoodthataparticulargraduateoftheschoolwillattendcollege,or
when you wonder whether to bring up the Tour de France in a conversation with a
Frenchmanyoujustmet.
Stereotypingisabadwordinourculture,butinmyusageitisneutral.Oneofthebasic
characteristics of System 1 is that it represents categories as norms and prototypical
exemplars.Thisishowwethinkofhorses,refrigerators,andNewYorkpoliceofficers;we
hold in memory a representation of one or more “normal” members of each of these
categories. When the categories are social, these representations are called stereotypes.
Some stereotypes are perniciously wrong, and hostile stereotyping can have dreadful
consequences,butthepsychologicalfactscannotbeavoided:stereotypes,bothcorrectand
false,arehowwethinkofcategories.
Youmaynotetheirony.Inthecontextofthecabproblem,theneglectofbase-rate
informationisacognitiveflaw,afailureofBayesianreasoning,andtherelianceoncausal
baseratesisdesirable.StereotypingtheGreendriversimprovestheaccuracyofjudgment.
In other contexts, however, such as hiring or profiling, there is a strong social norm
against stereotyping, which is also embedded in the law. This is as it should be. In
sensitivesocialcontexts,wedonotwanttodrawpossiblyerroneousconclusionsaboutthe
individualfromthestatisticsofthegroup.Weconsideritmorallydesirableforbaserates
to be treated as statistical facts about the group rather than as presumptive facts about
individuals.Inotherwords,werejectcausalbaserates.
Thesocialnormagainststereotyping,includingtheoppositiontoprofiling,hasbeen
highly beneficial in creating a more civilized and more equal society. It is useful to
remember, however, that neglecting valid stereotypes inevitably results in suboptimal
judgments.Resistancetostereotypingisalaudablemoralposition,butthesimplisticidea
that the resistance is costless is wrong. The costs are worth paying to achieve a better
society,butdenyingthatthecostsexist,whilesatisfyingtothesoulandpoliticallycorrect,
isnotscientificallydefensible.Reliance on theaffectheuristic is commoninpolitically
charged arguments. The positions we favor have no cost and those we oppose have no
benefits.Weshouldbeabletodobetter.
CausalSituations
Amos and I constructed the variants of the cab problem, but we did not invent the
powerfulnotionofcausalbaserates;weborroweditfromthepsychologistIcekAjzen.In
his experiment, Ajzen showed his participants brief vignettes describing some students
whohad takenan examat Yaleand askedthe participantsto judgethe probabilitythat
each student had passed the test. The manipulation of causal bs oase rates was
straightforward:Ajzentoldonegroupthatthestudentstheysawhadbeendrawnfroma
classinwhich75%passedtheexam,andtoldanothergroupthatthesamestudentshad
beeninaclassinwhichonly25%passed.Thisisapowerfulmanipulation,becausethe
baserateofpassingsuggeststheimmediateinferencethatthetestthatonly25%passed
musthavebeenbrutallydifficult.Thedifficultyofatestis,ofcourse,oneofthecausal
factorsthatdetermineeverystudent’soutcome.Asexpected,Ajzen’ssubjectswerehighly
sensitivetothecausalbaserates,andeverystudentwasjudgedmorelikelytopassinthe
high-successconditionthaninthehigh-failurerate.
Ajzenusedaningeniousmethodtosuggestanoncausalbaserate.Hetoldhissubjects
thatthestudentstheysawhadbeendrawnfromasample,whichitselfwasconstructedby
selectingstudentswhohadpassedorfailedtheexam.Forexample,theinformationforthe
high-failuregroupreadasfollows:
The investigator was mainly interested in the causes of failure and constructed a
sampleinwhich75%hadfailedtheexamination.
Note the difference. This base rate is a purely statistical fact about the ensemble from
whichcaseshavebeendrawn.Ithasnobearingonthequestionasked,whichiswhether
theindividualstudentpassedorfailedthetest.Asexpected,theexplicitlystatedbaserates
had some effects on judgment, but they had much less impact than the statistically
equivalent causal base rates. System 1 can deal with stories in which the elements are
causallylinked,butitisweakinstatisticalreasoning.ForaBayesianthinker,ofcourse,
theversionsareequivalent.Itistemptingtoconcludethatwehavereachedasatisfactory
conclusion:causalbaseratesareused;merelystatisticalfactsare(moreorless)neglected.
The next study, one of my all-time favorites, shows that the situation is rather more
complex.
CanPsychologybeTaught?
The reckless cabdrivers and the impossibly difficult exam illustrate two inferences that
people can draw from causal base rates: a stereotypical trait that is attributed to an
individual,andasignificantfeatureofthesituationthataffectsanindividual’soutcome.
The participants in the experiments made the correct inferences and their judgments
improved.Unfortunately,thingsdonotalwaysworkoutsowell.TheclassicexperimentI
describenextshowsthatpeoplewillnotdrawfrombase-rateinformationaninferencethat
conflicts with other beliefs. It also supports the uncomfortableconclusion that teaching
psychologyismostlyawasteoftime.
The experiment was conducted a long time ago by the social psychologistRichard
NisbettandhisstudentEugeneBorgida,attheUniversityofMichigan.Theytoldstudents
abouttherenowned“helpingexperiment”thathadbeenconductedafewyearsearlierat
NewYork University.Participants inthat experiment wereled toindividual booths and
invitedtospeakovertheintercomabouttheirpersonallivesandproblems.Theywereto
talkinturnforabouttwominutes.Onlyonemicrophonewasactiveatanyonetime.There
weresixparticipantsineachgroup,oneofwhomwasastooge.Thestoogespokefirst,
followingascriptpreparedbytheexperimenters.Hedescribedhisproblemsadjustingto
New York and admitted with obvious embarrassment that he was prone to seizures,
especiallywhenstressed.Alltheparticipantsthenhadaturn.Whenthemicrophonewas
againturnedovertothestooge,hebecameagitatedandincoherent,saidhefeltaseizure
comingon,andpeoaskedforsomeonetohelphim.Thelastwordsheardfromhimwere,
“C-could somebody-er-er-help-er-uh-uh-uh [choking sounds]. I…I’m gonna die-er-er-er
I’m…gonnadie-er-er-IseizureI-er[chokes,thenquiet].”Atthispointthemicrophoneof
thenext participantautomaticallybecame active, andnothing morewas heardfrom the
possiblydyingindividual.
Whatdoyouthinktheparticipantsintheexperimentdid?Sofarastheparticipants
knew, one of them was having a seizure and had asked for help. However, there were
severalotherpeoplewhocouldpossiblyrespond,soperhapsonecouldstaysafelyinone’s
booth.Theseweretheresults:onlyfourofthefifteenparticipantsrespondedimmediately
totheappealforhelp.Sixnevergotoutoftheirbooth,andfiveotherscameoutonlywell
afterthe“seizurevictim”apparentlychoked.Theexperimentshowsthatindividualsfeel
relieved of responsibility when they know that others have heard the same request for
help.
Didtheresultssurpriseyou?Veryprobably.Mostofusthinkofourselvesasdecent
peoplewhowouldrushtohelpinsuchasituation,andweexpectotherdecentpeopleto
dothesame.Thepointoftheexperiment,ofcourse,wastoshowthatthisexpectationis
wrong.Evennormal,decentpeopledonotrushtohelpwhentheyexpectotherstotakeon
theunpleasantnessofdealingwithaseizure.Andthatmeansyou,too.
Areyouwillingtoendorsethefollowingstatement?“WhenIreadtheprocedureof
thehelping experiment Ithought Iwould cometo the stranger’s helpimmediately,as I
probablywouldifIfoundmyselfalonewithaseizurevictim.Iwasprobablywrong.IfI
findmyselfinasituationinwhichotherpeoplehaveanopportunitytohelp,Imightnot
step forward. The presence of others would reduce my sense of personal responsibility
morethanIinitiallythought.”Thisiswhatateacherofpsychologywouldhopeyouwould
learn.Wouldyouhavemadethesameinferencesbyyourself?
Thepsychologyprofessorwhodescribesthehelpingexperimentwantsthestudentsto
viewthelowbaserateascausal,justasinthecaseofthefictitiousYaleexam.Hewants
themtoinfer,inbothcases,thatasurprisinglyhighrateoffailureimpliesaverydifficult
test. The lesson students are meant to take away is that some potent feature of the
situation,suchasthediffusionofresponsibility,inducesnormalanddecentpeoplesuchas
themtobehaveinasurprisinglyunhelpfulway.
Changingone’smindabouthumannatureishardwork,andchangingone’smindfor
theworseaboutoneselfisevenharder.NisbettandBorgidasuspectedthatstudentswould
resisttheworkandtheunpleasantness.Ofcourse,thestudentswouldbeableandwilling
to recite the details of the helping experiment on a test, and would even repeat the
“official”interpretationintermsofdiffusionofresponsibility.Butdidtheirbeliefsabout
human nature really change? To find out, Nisbett and Borgida showed them videos of
brief interviews allegedly conducted with two people who had participated in the New
Yorkstudy.Theinterviewswereshortandbland.Theintervieweesappearedtobenice,
normal,decentpeople.Theydescribedtheirhobbies,theirspare-timeactivities,andtheir
plans for the future, which were entirely conventional. After watching the video of an
interview,thestudentsguessedhowquicklythatparticularpersonhadcometotheaidof
thestrickenstranger.
ToapplyBayesianreasoningtothetaskthestudentswereassigned,youshouldfirstask
yourselfwhatyouwouldhaveguessedabouttheastwoindividualsifyouhadnotseen
theirinterviews.Thisquestionisansweredbyconsultingthebaserate.Wehavebeentold
thatonly4ofthe15participantsintheexperimentrushedtohelpafterthefirstrequest.
Theprobabilitythatanunidentifiedparticipanthadbeenimmediatelyhelpfulistherefore
27%.Thusyourpriorbeliefaboutanyunspecifiedparticipantshouldbethathedidnot
rushto help. Next,Bayesian logic requiresyou to adjustyour judgment inlight of any
relevantinformationabouttheindividual.However,thevideoswerecarefullydesignedto
beuninformative;theyprovidednoreasontosuspectthattheindividualswouldbeeither
more or less helpful than a randomly chosen student. In the absence of useful new
information,theBayesiansolutionistostaywiththebaserates.
NisbettandBorgidaaskedtwogroupsofstudentstowatchthevideosandpredictthe
behaviorofthetwoindividuals.Thestudentsinthefirstgroupweretoldonlyaboutthe
procedureofthehelpingexperiment,notaboutitsresults.Theirpredictionsreflectedtheir
viewsofhumannatureandtheirunderstandingofthesituation.Asyoumightexpect,they
predicted that both individuals would immediately rush to the victim’s aid. The second
group of students knew both the procedure of the experiment and its results. The
comparison of the predictions of the two groups provides an answer to a significant
question: Did students learn from the results of the helping experiment anything that
significantlychangedtheirwayofthinking?Theanswerisstraightforward:theylearned
nothingatall.Theirpredictionsaboutthetwoindividualswereindistinguishablefromthe
predictions made by students who had not been exposed to the statistical results of the
experiment.Theyknewthebaserateinthegroup from which theindividualshadbeen
drawn,buttheyremainedconvincedthatthepeopletheysawonthevideohadbeenquick
tohelpthestrickenstranger.
Forteachersofpsychology,theimplicationsofthisstudyaredisheartening.Whenwe
teachourstudentsaboutthebehaviorofpeopleinthehelpingexperiment,weexpectthem
tolearnsomethingtheyhadnotknownbefore;wewishtochangehowtheythinkabout
people’sbehaviorinaparticularsituation.ThisgoalwasnotaccomplishedintheNisbett-
Borgidastudy,andthereisnoreasontobelievethattheresultswouldhavebeendifferent
if they had chosen another surprising psychological experiment. Indeed, Nisbett and
Borgidareportedsimilarfindingsinteachinganotherstudy,inwhichmildsocialpressure
caused people to accept much more painful electric shocks than most of us (and them)
wouldhaveexpected.Studentswhodonotdevelopanewappreciationforthepowerof
social setting have learned nothing of value from the experiment. The predictions they
make about random strangers, or about their own behavior, indicate that they have not
changedtheirviewofhowtheywouldhavebehaved.InthewordsofNisbettandBorgida,
students “quietly exempt themselves” (and their friends and acquaintances) from the
conclusionsofexperimentsthatsurprisethem.Teachersofpsychologyshouldnotdespair,
however,becauseNisbettandBorgidareportawaytomaketheirstudentsappreciatethe
pointofthehelpingexperiment.Theytookanewgroupofstudentsandtaughtthemthe
procedureoftheexperimentbutdidnottellthemthegroupresults.Theyshowedthetwo
videosandsimplytoldtheirstudentsthatthetwoindividualstheyhadjustseenhadnot
helped the stranger, then asked them to guess the global results. The outcome was
dramatic:thestudents’guesseswereextremelyaccurate.
Toteachstudentsanypsychologytheydidnotknowbefore,youmustsurprisethem.
But which surprise will do? Nisbett and Borgida found that when they presented their
studentswithasurprisingstatisticisalfact,thestudentsmanagedtolearnnothingatall.
Butwhenthestudentsweresurprisedbyindividualcases—twonicepeoplewhohadnot
helped—they immediately made the generalization and inferred that helping is more
difficultthantheyhadthought.NisbettandBorgidasummarizetheresultsinamemorable
sentence:
Subjects’unwillingnesstodeducetheparticularfromthegeneralwasmatchedonly
bytheirwillingnesstoinferthegeneralfromtheparticular.
This is a profoundly important conclusion. People who are taught surprising statistical
factsabouthumanbehaviormaybeimpressedtothepointoftellingtheirfriendsabout
whattheyhave heard,but this doesnot mean thattheir understanding ofthe worldhas
really changed. The test of learning psychology is whether your understanding of
situationsyouencounterhaschanged,notwhetheryouhavelearnedanewfact.Thereisa
deep gap between our thinking about statistics and our thinking about individual cases.
Statisticalresultswithacausalinterpretationhaveastrongereffectonourthinkingthan
noncausal information. But even compelling causal statistics will not change long-held
beliefsorbeliefsrootedinpersonalexperience.Ontheotherhand,surprisingindividual
cases have a powerful impact and are a more effective tool for teaching psychology
becausetheincongruitymustberesolvedandembeddedinacausalstory.Thatiswhythis
bookcontainsquestionsthatareaddressedpersonallytothereader.Youaremorelikelyto
learnsomethingbyfindingsurprisesinyourownbehaviorthanbyhearingsurprisingfacts
aboutpeopleingeneral.
SpeakingofCausesandStatistics
“Wecan’tassumethattheywillreallylearnanythingfrommerestatistics.Let’sshow
themoneortworepresentativeindividualcasestoinfluencetheirSystem1.”
“Noneedtoworryaboutthisstatisticalinformationbeingignored.Onthecontrary,it
willimmediatelybeusedtofeedastereotype.”
P
RegressiontotheMean
I had one of the most satisfying eureka experiences of my career while teaching flight
instructorsintheIsraeliAirForceaboutthepsychologyofeffectivetraining.Iwastelling
them about an important principle of skill training: rewards for improved performance
workbetterthanpunishmentofmistakes.Thispropositionissupportedbymuchevidence
fromresearchonpigeons,rats,humans,andotheranimals.
WhenIfinishedmyenthusiasticspeech,oneofthemostseasonedinstructorsinthe
groupraisedhishandandmadeashortspeechofhisown.Hebeganbyconcedingthat
rewardingimprovedperformancemightbegoodforthebirds,buthedeniedthatitwas
optimalforflightcadets.Thisiswhathesaid:“OnmanyoccasionsIhavepraisedflight
cadetsforcleanexecutionofsomeaerobaticmaneuver.Thenexttimetheytrythesame
maneuvertheyusuallydoworse.Ontheotherhand,Ihaveoftenscreamedintoacadet’s
earphoneforbadexecution,andingeneralhedoesbetterttaskyryabrtworeponhisnext
try. So please don’t tell us that reward works and punishment does not, because the
oppositeisthecase.”
This was a joyous moment of insight, when I saw in a new light a principle of
statistics that I had been teaching for years. The instructor was right—but he was also
completelywrong!Hisobservationwasastuteandcorrect:occasionsonwhichhepraised
a performance were likely to be followed by a disappointing performance, and
punishmentsweretypicallyfollowedbyanimprovement.Buttheinferencehehaddrawn
abouttheefficacyofrewardandpunishmentwascompletelyoffthemark.Whathehad
observed is known as regression to the mean, which in that case was due to random
fluctuations in the quality of performance. Naturally, he praised only a cadet whose
performancewasfar better thanaverage.Butthe cadetwasprobably just luckyonthat
particularattemptandthereforelikelytodeteriorateregardlessofwhetherornothewas
praised. Similarly, the instructor would shout into a cadet’s earphones only when the
cadet’sperformancewasunusuallybadandthereforelikelytoimproveregardlessofwhat
the instructor did. The instructor had attached a causal interpretation to the inevitable
fluctuationsofarandomprocess.
Thechallengecalledforaresponse,butalessoninthealgebraofpredictionwould
notbeenthusiasticallyreceived.Instead,Iusedchalktomarkatargetonthefloor.Iasked
every officer in the room to turn his back to the target and throw two coins at it in
immediate succession, without looking. We measured the distances from the target and
wrotethetworesultsofeachcontestantontheblackboard.Thenwerewrotetheresultsin
order,fromthebesttotheworstperformanceonthefirsttry.Itwasapparentthatmost
(butnotall)ofthosewhohaddonebestthefirsttimedeterioratedontheirsecondtry,and
thosewhohaddonepoorlyonthefirstattemptgenerallyimproved.Ipointedouttothe
instructorsthatwhattheysawontheboardcoincidedwithwhatwehadheardaboutthe
performance of aerobatic maneuvers on successive attempts: poor performance was
typicallyfollowedbyimprovementandgoodperformancebydeterioration,withoutany
helpfromeitherpraiseorpunishment.
ThediscoveryImadeonthatdaywasthattheflightinstructorsweretrappedinan
unfortunatecontingency:becausetheypunishedcadetswhenperformancewaspoor,they
were mostly rewarded by a subsequent improvement, even if punishment was actually
ineffective. Furthermore, the instructors were not alone in that predicament. I had
stumbled onto a significant fact of the human condition: the feedback to which life
exposesusisperverse.Becausewetendtobenicetootherpeoplewhentheypleaseus
andnastywhentheydonot,wearestatisticallypunishedforbeingniceandrewardedfor
beingnasty.
TalentandLuck
Afewyearsago,JohnBrockman,whoeditstheonlinemagazineEdge,askedanumberof
scientiststoreporttheir“favoriteequation.”Theseweremyofferings:
success=talent+luck
greatsuccess=alittlemoretalent+alotofluck
Theunsurprisingideathatluckoftencontributestosuccesshassurprisingconsequences
when we apply it to the first two days of a high-level golf tournament. To keep things
simple,assumethatonbothdaystheaveragescoreofthecompetitorswasatpar72.We
focusonaplayerwhodidveryedwellonthefirstday,closingwithascoreof66.What
canwelearnfromthatexcellentscore?Animmediateinferenceisthatthegolferismore
talentedthantheaverageparticipantinthetournament.Theformulaforsuccesssuggests
thatanotherinferenceisequallyjustified:thegolferwhodidsowellonday1probably
enjoyed better-than-average luck on that day. If you accept that talent and luck both
contributetosuccess,theconclusionthatthesuccessfulgolferwasluckyisaswarranted
astheconclusionthatheistalented.
Bythesametoken,ifyoufocusonaplayerwhoscored5overparonthatday,you
havereasontoinferboththatheisratherweakandhadabadday.Ofcourse,youknow
thatneitheroftheseinferencesiscertain.Itisentirelypossiblethattheplayerwhoscored
77isactuallyverytalentedbuthadanexceptionallydreadfulday.Uncertainthoughthey
are, the following inferences from the score on day 1 are plausible and will be correct
moreoftenthantheyarewrong.
above-averagescoreonday1=above-averagetalent+luckyonday1
and
below-averagescoreonday1=below-averagetalent+unluckyonday1
Now,supposeyouknowagolfer’sscoreonday1andareaskedtopredicthisscore
onday2.Youexpectthegolfertoretainthesameleveloftalentonthesecondday,so
yourbestguesseswillbe“aboveaverage”forthefirstplayerand“belowaverage”forthe
secondplayer.Luck,ofcourse,isadifferentmatter.Sinceyouhavenowayofpredicting
the golfers’ luck on the second (or any) day, your best guess must be that it will be
average,neithergoodnorbad.Thismeansthatintheabsenceofanyotherinformation,
your best guess about the players’ score on day 2 should not be a repeat of their
performanceonday1.Thisisthemostyoucansay:
Thegolferwhodidwellonday1islikelytobesuccessfulonday2aswell,butless
thanonthefirst,becausetheunusualluckheprobablyenjoyedonday1isunlikelyto
hold.
Thegolferwhodidpoorlyonday1willprobablybebelowaverageonday2,but
willimprove,becausehisprobablestreakofbadluckisnotlikelytocontinue.
We also expect the difference between the two golfers to shrink on the second day,
althoughourbestguessisthatthefirstplayerwillstilldobetterthanthesecond.
Mystudentswerealwayssurprisedtohearthatthebestpredictedperformanceonday
2ismoremoderate,closertotheaveragethantheevidenceonwhichitisbased(thescore
onday1).Thisiswhythepatterniscalledregressiontothemean.Themoreextremethe
originalscore,themoreregressionweexpect,becauseanextremelygoodscoresuggestsa
veryluckyday.Theregressivepredictionisreasonable,butitsaccuracyisnotguaranteed.
Afewofthegolferswhoscored66onday1willdoevenbetteronthesecondday,iftheir
luckimproves.Mostwilldoworse,becausetheirluckwillnolongerbeaboveaverage.
Nowletus go againstthetimearrow.Arrangetheplayersbytheirperformanceon
day2andlookattheirperformanceonday1.Youwillfindpreciselythesamepatternof
regressiontothemean.Thegolferswhodidbestonday2wereprobablyluckyonthat
day,andthebestguessisthattheyhadbeenlessluckyandhaddonefilesswellonday1.
Thefactthatyouobserveregressionwhenyoupredictanearlyeventfromalaterevent
shouldhelpconvinceyouthatregressiondoesnothaveacausalexplanation.
Regression effects are ubiquitous, and so are misguided causal stories to explain
them. A well-known example is the “Sports Illustrated jinx,” the claim that an athlete
whose picture appears on the cover of the magazine is doomed to perform poorly the
followingseason.Overconfidenceandthepressureofmeetinghighexpectationsareoften
offeredasexplanations.Butthereisasimpleraccountofthejinx:anathletewhogetsto
be on the cover of Sports Illustrated must have performed exceptionally well in the
precedingseason,probablywiththeassistanceofanudgefromluck—andluckisfickle.
Ihappenedtowatchthemen’sskijumpeventintheWinterOlympicswhileAmos
andIwerewritinganarticleaboutintuitiveprediction.Eachathletehastwojumpsinthe
event, and the results are combined for the final score. I was startled to hear the
sportscaster’s comments while athletes were preparing for their second jump: “Norway
hadagreatfirstjump;hewillbetense,hopingtoprotecthisleadandwillprobablydo
worse”or“Swedenhadabadfirstjumpandnowheknowshehasnothingtoloseandwill
berelaxed,whichshouldhelphimdobetter.”Thecommentatorhadobviouslydetected
regressiontothemeanandhadinventedacausalstoryforwhichtherewasnoevidence.
Thestoryitselfcouldevenbetrue.Perhapsifwemeasuredtheathletes’pulsebeforeeach
jumpwemightfindthattheyareindeedmorerelaxedafterabadfirstjump.Andperhaps
not.Thepointtorememberisthatthechangefromthefirsttothesecondjumpdoesnot
needacausalexplanation.Itisamathematicallyinevitableconsequenceofthefactthat
luck played a role in the outcome of the first jump. Not a very satisfactory story—we
wouldallpreferacausalaccount—butthatisallthereis.
UnderstandingRegression
Whetherundetectedorwronglyexplained,thephenomenonofregressionisstrangetothe
humanmind.Sostrange,indeed,thatitwasfirstidentifiedandunderstoodtwohundred
yearsafterthetheoryofgravitationanddifferentialcalculus.Furthermore,ittookoneof
the best minds of nineteenth-century Britain to make sense of it, and that with great
difficulty.
Regressiontothemeanwasdiscoveredandnamedlateinthenineteenthcenturyby
SirFrancisGalton,ahalfcousinofCharlesDarwinandarenownedpolymath.Youcan
sensethethrillofdiscoveryinanarticlehepublishedin1886underthetitle“Regression
towards Mediocrity in Hereditary Stature,” which reports measurements of size in
successivegenerationsofseedsandincomparisonsoftheheightofchildrentotheheight
oftheirparents.Hewritesabouthisstudiesofseeds:
Theyyieldedresultsthatseemedverynoteworthy,andIusedthemasthebasisofa
lecture before the Royal Institution on February 9th, 1877. It appeared from these
experimentsthattheoffspringdidnottendtoresembletheirparentseedsinsize,but
tobealwaysmoremediocrethanthey—tobesmallerthantheparents,iftheparents
were large; to be larger than the parents, if the parents were very small…The
experimentsshowed further thatthe meanfilial regression towardsmediocrity was
directlyproportionaltotheparentaldeviationfromit.
Galton obviously expected his learned audience at the Royal Institution—the oldest
independent research society in the world—to be as surprised by his “noteworthy
observation” as he had been. What is truly noteworthy is that he was surprised by a
statisticalregularity thatis as common as theair we breathe. Regression effects can be
foundwhereverwelook,butwedonotrecognizethemforwhattheyare.Theyhidein
plain sight. It took Galton several years to work his way from his discovery of filial
regression in size to the broader notion that regression inevitably occurs when the
correlationbetweentwomeasuresislessthanperfect,andheneededthehelpofthemost
brilliantstatisticiansofhistimetoreachthatconclusion.
OneofthehurdlesGaltonhadtoovercomewastheproblemofmeasuringregression
betweenvariablesthataremeasuredondifferentscales,suchasweightandpianoplaying.
Thisisdonebyusingthepopulationasastandardofreference.Imaginethatweightand
pianoplayinghavebeenmeasuredfor100childreninallgradesofanelementaryschool,
andthattheyhavebeenrankedfromhightolowoneachmeasure.IfJaneranksthirdin
piano playing and twenty-seventh in weight, it is appropriate to say that she is a better
pianistthansheistall.Letusmakesomeassumptionsthatwillsimplifythings:
Atanyage,
Piano-playingsuccessdependsonlyonweeklyhoursofpractice.
Weightdependsonlyonconsumptionoficecream.
Icecreamconsumptionandweeklyhoursofpracticeareunrelated.
Now, using ranks (or the standard scores that statisticians prefer), we can write some
equations:
weight=age+icecreamconsumption
pianoplaying=age+weeklyhoursofpractice
Youcanseethattherewillberegressiontothemeanwhenwepredictpianoplayingfrom
weight,orviceversa.IfallyouknowaboutTomisthatherankstwelfthinweight(well
aboveaverage),youcaninfer(statistically)thatheisprobablyolderthanaverageandalso
that he probably consumes more ice cream than other children. If all you know about
Barbaraisthatsheiseighty-fifthinpiano(farbelowtheaverageofthegroup),youcan
inferthatsheislikelytobeyoungandthatsheislikelytopracticelessthanmostother
children.
Thecorrelationcoefficientbetweentwomeasures,whichvariesbetween0and1,isa
measureoftherelativeweightofthefactorstheyshare.Forexample,weallsharehalfour
genes with each of our parents, and for traits in which environmental factors have
relativelylittleinfluence,suchasheight,thecorrelationbetweenparentandchildisnot
farfrom.50.Toappreciatethemeaningofthecorrelationmeasure,thefollowingaresome
examplesofcoefficients:
ThecorrelationbetweenthesizeofobjectsmeasuredwithprecisioninEnglishorin
metricunitsis1.Anyfactorthatinfluencesonemeasurealsoinfluencestheother;
100%ofdeterminantsareshared.
The correlation between self-reported height and weight among adult American
males is .41. If you included women and children, the correlation would be much
higher,becauseindividuals’genderandageinfluenceboththeirheightannwdtheir
weight,boostingtherelativeweightofsharedfactors.
ThecorrelationbetweenSATscoresandcollegeGPAisapproximately.60.However,
thecorrelationbetweenaptitudetestsandsuccessingraduateschoolismuchlower,
largelybecausemeasuredaptitudevarieslittleinthisselectedgroup.Ifeveryonehas
similar aptitude, differences in this measure are unlikely to play a large role in
measuresofsuccess.
The correlation between income and education level in the United States is
approximately.40.
Thecorrelationbetweenfamilyincomeandthelastfourdigitsoftheirphonenumber
is0.
IttookFrancisGaltonseveralyearstofigureoutthatcorrelationandregressionare
nottwoconcepts—theyaredifferentperspectivesonthesameconcept.Thegeneralruleis
straightforward but has surprising consequences: whenever the correlation between two
scoresisimperfect,therewillberegressiontothemean.ToillustrateGalton’sinsight,take
apropositionthatmostpeoplefindquiteinteresting:
Highlyintelligentwomentendtomarrymenwhoarelessintelligentthantheyare.
Youcangetagoodconversationstartedatapartybyaskingforanexplanation,andyour
friends will readily oblige. Even people who have had some exposure to statistics will
spontaneously interpret the statement in causal terms. Some may think of highly
intelligentwomenwantingtoavoidthecompetitionofequallyintelligentmen,orbeing
forced to compromise in their choice of spouse because intelligent men do not want to
competewithintelligentwomen.More far-fetched explanationswillcomeupat agood
party.Nowconsiderthisstatement:
Thecorrelationbetweentheintelligencescoresofspousesislessthanperfect.
This statement is obviously true and not interesting at all. Who would expect the
correlation to be perfect? There is nothing to explain. But the statement you found
interesting and the statement you found trivial are algebraically equivalent. If the
correlationbetweentheintelligenceofspousesislessthanperfect(andifmenandwomen
onaveragedonotdifferinintelligence),thenitisamathematicalinevitabilitythathighly
intelligentwomen will be marriedto husbandswho areon averageless intelligentthan
theyare(andviceversa,ofcourse).Theobservedregressiontothemeancannotbemore
interestingormoreexplainablethantheimperfectcorrelation.
You probably sympathize with Galton’s struggle with the concept of regression.
Indeed,thestatisticianDavidFreedmanusedtosaythatifthetopicofregressioncomes
upinacriminalorciviltrial,thesidethatmustexplainregressiontothejurywilllosethe
case.Why isit sohard? Themain reasonfor thedifficultyisa recurrenttheme ofthis
book:ourmindisstronglybiasedtowardcausalexplanationsanddoesnotdealwellwith
“merestatistics.”Whenourattentioniscalledtoanevent,associativememorywilllook
for its cause—more precisely, activation will automatically spread to any cause that is
already stored in memory. Causal explanations will be evoked when regression is
detected,buttheywillbewrongbecausethetruthisthatregressiontothemeanhasan
explanationbutdoesnothaveacause.Theeventthatattractsourattentioninthegolfing
tournament is the frequent deterioration of the performance of the golfers who werecte
successfulonday1.Thebestexplanationofitisthatthosegolferswereunusuallylucky
thatday,butthisexplanationlacksthecausalforcethatourmindsprefer.Indeed,wepay
people quite well to provide interesting explanations of regression effects. A business
commentator who correctly announces that “the business did better this year because it
haddonepoorlylastyear”islikelytohaveashorttenureontheair.
OurdifficultieswiththeconceptofregressionoriginatewithbothSystem1andSystem2.
Withoutspecialinstruction,andinquiteafewcasesevenaftersomestatisticalinstruction,
the relationship between correlation and regression remains obscure. System 2 finds it
difficult to understand and learn. This is due in part to the insistent demand for causal
interpretations,whichisafeatureofSystem1.
Depressedchildrentreatedwithanenergydrinkimprovesignificantlyoverathree-
monthperiod.
Imadeupthisnewspaperheadline,butthefactitreportsistrue:ifyoutreatedagroupof
depressed children for some time with an energy drink, they would show a clinically
significantimprovement.Itisalsothecasethatdepressedchildrenwhospendsometime
standingontheirheadorhugacatfortwentyminutesadaywillalsoshowimprovement.
Mostreaders of such headlines willautomatically inferthat the energydrink or the cat
huggingcausedanimprovement,butthisconclusioniscompletelyunjustified.Depressed
children are an extreme group, they are more depressed than most other children—and
extremegroupsregresstothemeanovertime.Thecorrelationbetweendepressionscores
onsuccessiveoccasionsoftestingislessthanperfect,sotherewillberegressiontothe
mean:depressedchildrenwillgetsomewhatbetterovertimeeveniftheyhugnocatsand
drinknoRedBull.Inordertoconcludethatanenergydrink—oranyothertreatment—is
effective,youmustcompareagroupofpatientswhoreceivethistreatmenttoa“control
group” that receives no treatment (or, better, receives a placebo). The control group is
expectedtoimprovebyregressionalone,andtheaimoftheexperimentistodetermine
whetherthetreatedpatientsimprovemorethanregressioncanexplain.
Incorrectcausalinterpretationsofregressioneffectsarenotrestrictedtoreadersofthe
popular press. The statistician Howard Wainer has drawn up a long list of eminent
researcherswhohavemadethesamemistake—confusingmerecorrelationwithcausation.
Regressioneffectsareacommonsourceoftroubleinresearch,andexperiencedscientists
developahealthyfearofthetrapofunwarrantedcausalinference.
One of my favorite examples of the errors of intuitive prediction is adapted from Max
Bazerman’sexcellenttextJudgmentinManagerialDecisionMaking:
Youarethesalesforecasterforadepartmentstorechain.Allstoresaresimilarinsize
andmerchandiseselection,buttheirsalesdifferbecauseoflocation,competition,and
random factors. You are given the results for 2011 and asked to forecast sales for
2012.Youhavebeeninstructedtoaccepttheoverallforecastofeconomiststhatsales
willincreaseoverallby10%.Howwouldyoucompletethefollowingtable?
Store 2011 2012
1 $11,000,000 ________
2 $23,000,000 ________
3 $18,000,000 ________
4 $29,000,000 ________
Total $61,000,000 $67,100,000
Havingreadthischapter,youknowthattheobvioussolutionofadding10%tothe
sales of each store is wrong. You want your forecasts to be regressive, which requires
adding more than 10% to the low-performing branches and adding less (or even
subtracting)toothers.Butifyouaskotherpeople,youarelikelytoencounterpuzzlement:
Whydoyoubotherthemwithanobviousquestion?AsGaltonpainfullydiscovered,the
conceptofregressionisfarfromobvious.
SpeakingofRegressiontoMediocrity
“Shesaysexperiencehastaughtherthatcriticismismoreeffectivethanpraise.What
shedoesn’tunderstandisthatit’sallduetoregressiontothemean.”
“Perhapshissecondinterviewwaslessimpressivethanthefirstbecausehewas
afraidofdisappointingus,butmorelikelyitwashisfirstthatwasunusuallygood.”
“Ourscreeningprocedureisgoodbutnotperfect,soweshouldanticipateregression.
We shouldn’t be surprised that the very best candidates often fail to meet our
expectations.”
P
TamingIntuitivePredictions
Life presents us with many occasions to forecast. Economists forecast inflation and
unemployment, financial analysts forecast earnings, military experts predict casualties,
venture capitalists assess profitability, publishers and producers predict audiences,
contractorsestimatethetimerequiredtocompleteprojects,chefsanticipatethedemand
for the dishes on their menu, engineers estimate the amount of concrete needed for a
building,firegroundcommandersassessthenumberoftrucksthatwillbeneededtoput
outafire.Inourprivatelives,weforecastourspouse’sreactiontoaproposedmoveorour
ownfutureadjustmenttoanewjob.
Somepredictivejudgments,suchasthosemadebyengineers,relylargelyonlook-up
tables, precise calculations, and explicit analyses of outcomes observed on similar
occasions.OthersinvolveintuitionandSystem1,intwomainvarieties.Someintuitions
draw primarily on skill and expertise acquired by repeated experience. The rapid and
automatic judgments and choices of chess masters, fireground commanders, and
physicians that Gary Klein has described in Sources of Power and elsewhere illustrate
theseskilledintuitions,inwhichasolutiontothecurrentproblemcomestomindquickly
becausefamiliarcuesarerecognized.
Other intuitions, which are sometimes subjectively indistinguishable from the first,
arisefromtheoperationofheuristicsthatoftensubstituteaneasyquestionfortheharder
onethatwasasked.Intuitivejudgmentscanbemadewithhighconfidenceevenwhenthey
arebased onnonregressive assessmentsof weak evidence. Of course,many judgments,
especially in the professional domain, are influenced by a combination of analysis and
intuition.
NonregressiveIntuitions
Letusreturntoapersonwehavealreadymet:
Julieiscurrentlyaseniorinastateuniversity.Shereadfluentlywhenshewasfour
yearsold.Whatishergradepointaverage(GPA)?
People who are familiar with the American educational scene quickly come up with a
number, which is often in the vicinity of 3.7 or 3.8. How does this occur? Several
operationsofSystem1areinvolved.
Acausallinkbetweentheevidence(Julie’sreading)andthetargetoftheprediction
(herGPA)issought.Thelinkcanbeindirect.Inthisinstance,earlyreadinganda
highGDParebothindicationsofacademictalent.Someconnectionisnecessary.You
(yourSystem2)wouldprobablyrejectasirrelevantareportofJuliewinningafly
fishing competitiowhired D=n or excelling at weight lifting in high school. The
process is effectively dichotomous. We are capable of rejecting information as
irrelevant or false, but adjusting for smaller weaknesses in the evidence is not
something that System 1 can do. As a result, intuitive predictions are almost
completelyinsensitivetotheactualpredictivequalityoftheevidence.Whenalinkis
found, as in the case of Julie’s early reading, WY SIATI applies: your associative
memory quickly and automatically constructs the best possible story from the
informationavailable.
Next,theevidenceisevaluatedinrelationtoarelevantnorm.Howprecociousisa
child who reads fluently at age four? What relative rank or percentile score
correspondstothisachievement?Thegrouptowhichthechildiscompared(wecall
itareferencegroup)isnotfullyspecified,butthisisalsotheruleinnormalspeech:if
someonegraduating from collegeis describedas “quiteclever”you rarelyneed to
ask,“Whenyousay‘quiteclever,’whichreferencegroupdoyouhaveinmind?”
The next step involves substitution and intensity matching. The evaluation of the
flimsyevidence of cognitiveabilityinchildhoodissubstitutedasananswertothe
questionabouthercollegeGPA.Juliewillbeassignedthesamepercentilescorefor
herGPAandforherachievementsasanearlyreader.
The question specified that the answer must be on the GPA scale, which requires
anotherintensity-matchingoperation,fromageneralimpressionofJulie’sacademic
achievementstotheGPAthatmatchestheevidenceforhertalent.Thefinalstepisa
translation,fromanimpressionofJulie’srelativeacademicstandingtotheGPAthat
correspondstoit.
Intensity matching yields predictions that are as extreme as the evidence on which
theyarebased,leadingpeopletogivethesameanswertotwoquitedifferentquestions:
WhatisJulie’spercentilescoreonreadingprecocity?
WhatisJulie’spercentilescoreonGPA?
BynowyoushouldeasilyrecognizethatalltheseoperationsarefeaturesofSystem1.
Ilistedthemhereasanorderlysequenceofsteps,butofcoursethespreadofactivationin
associativememorydoesnotworkthisway.Youshouldimagineaprocessofspreading
activation that is initially prompted by the evidence and the question, feeds back upon
itself,andeventuallysettlesonthemostcoherentsolutionpossible.
AmosandIonceaskedparticipantsinanexperimenttojudgedescriptionsofeightcollege
freshmen,allegedlywrittenbyacounseloronthebasisofinterviewsoftheenteringclass.
Eachdescriptionconsistedoffiveadjectives,asinthefollowingexample:
intelligent,self-confident,well-read,hardworking,inquisitive
Weaskedsomeparticipantstoanswertwoquestions:
Howmuchdoesthisdescriptionimpressyouwithrespecttoacademicability?
What percentage of descriptions of freshmen do you believe would impress you
more?
Thequestionsrequireyoutoevaluatetheevidencebycomparingthedescriptionto
yournormfordescriptionsofstudentsbycounselors.Theveryexistenceofsuchanormis
remarkable.Althoughyousurelydonotknowhowyouacquiredit,youhaveafairlyclear
senseofhowmuchenthusiasmthedescriptionconveys:thecounselorbelievesthatthis
student is good, but not spectacularly good. There is room for stronger adjectives than
intelligent(brilliant,creative),well-read(scholarly,erudite,impressivelyknowledgeable),
andhardworking(passionate,perfectionist).Theverdict:verylikelytobeinthetop15%
butunlikelytobeinthetop3%.Thereisimpressiveconsensusinsuchjudgments,atleast
withinaculture.
Theotherparticipantsinourexperimentwereaskeddifferentquestions:
Whatisyourestimateofthegradepointaveragethatthestudentwillobtain?
WhatisthepercentageoffreshmenwhoobtainahigherGPA?
You need another look to detect the subtle difference between the two sets of
questions.Thedifferenceshouldbeobvious,butitisnot.Unlikethefirstquestions,which
required you only to evaluate the evidence, the second set involves a great deal of
uncertainty. The question refers to actual performance at the end of the freshman year.
Whathappenedduringtheyearsincetheinterviewwasperformed?Howaccuratelycan
you predict the student’s actual achievements in the first year at college from five
adjectives?WouldthecounselorherselfbeperfectlyaccurateifshepredictedGPAfrom
aninterview?
The objective of this study was to compare the percentile judgments that the
participants made when evaluating the evidence in one case, and when predicting the
ultimate outcome in another. The results are easy to summarize: the judgments were
identical.Althoughthetwosetsofquestionsdiffer(oneisaboutthedescription,theother
aboutthestudent’sfutureacademicperformance),theparticipantstreatedthemasifthey
werethesame.AswasthecasewithJulie,thepredictionofthefutureisnotdistinguished
froman evaluationof current evidence—predictionmatches evaluation. Thisis perhaps
thebestevidencewehavefortheroleofsubstitution.Peopleareaskedforapredictionbut
they substitute an evaluation of the evidence, without noticing that the question they
answerisnottheonetheywereasked.Thisprocessisguaranteedtogeneratepredictions
thataresystematicallybiased;theycompletelyignoreregressiontothemean.
DuringmymilitaryserviceintheIsraeliDefenseForces,Ispentsometimeattached
toaunitthatselectedcandidatesforofficertrainingonthebasisofaseriesofinterviews
andfieldtests.Thedesignatedcriterionforsuccessfulpredictionwasacadet’sfinalgrade
inofficerschool.Thevalidityoftheratingswasknowntoberatherpoor(Iwilltellmore
aboutitinalaterchapter).Theunitstillexistedyearslater,whenIwasaprofessorand
collaboratingwithAmosinthestudyofintuitivejudgment.Ihadgoodcontactswiththe
peopleattheunitandaskedthemforafavor.Inadditiontotheusualgradingsystemthey
usedtoevaluatethecandidates,Iaskedfortheirbestguessofthegradethateachofthe
futurecadetswouldobtaininofficerschool.Theycollectedafewhundredsuchforecasts.
Theofficerswhohadproducedtheprediofрctionswereallfamiliarwiththelettergrading
systemthattheschoolappliedtoitscadetsandtheapproximateproportionsofA’s,B’s,
etc.,amongthem.Theresultswerestriking:therelativefrequencyofA’sandB’sinthe
predictionswasalmostidenticaltothefrequenciesinthefinalgradesoftheschool.
These findings provide a compelling example of both substitution and intensity
matching. The officers who provided the predictions completely failed to discriminate
betweentwotasks:
theirusualmission,whichwastoevaluatetheperformanceofcandidatesduringtheir
stayattheunit
the task I had asked them to perform, which was an actual prediction of a future
grade
They had simply translated their own grades onto the scale used in officer school,
applying intensity matching. Once again, the failure to address the (considerable)
uncertainty of their predictions had led them to predictions that were completely
nonregressive.
ACorrectionforIntuitivePredictions
BacktoJulie,ourprecociousreader.ThecorrectwaytopredictherGPAwasintroduced
inthepreceding chapter. AsI did thereforgolf onsuccessivedays and forweightand
pianoplaying,Iwriteaschematicformulaforthefactorsthatdeterminereadingageand
collegegrades:
readingage=sharedfactors+factorsspecifictoreadingage=100%
GPA=sharedfactors+factorsspecifictoGPA=100%
Thesharedfactorsinvolvegeneticallydeterminedaptitude,thedegreetowhichthefamily
supports academic interests, and anything else that would cause the same people to be
precocious readers as children and academically successful as young adults. Of course
thereare manyfactors thatwould affectone ofthese outcomesand notthe other.Julie
couldhavebeenpushedtoreadearlybyoverlyambitiousparents,shemayhavehadan
unhappy love affair that depressed her college grades, she could have had a skiing
accidentduringadolescencethatleftherslightlyimpaired,andsoon.
Recall that the correlation between two measures—in the present case reading age
andGPA—isequaltotheproportionofsharedfactorsamongtheirdeterminants.Whatis
yourbestguessaboutthatproportion?Mymostoptimisticguessisabout30%.Assuming
this estimate, we have all we need to produce an unbiased prediction. Here are the
directionsforhowtogetthereinfoursimplesteps:
1. StartwithanestimateofaverageGPA.
2. DeterminetheGPAthatmatchesyourimpressionoftheevidence.
3. EstimatethecorrelationbetweenyourevidenceandGPA.
4. Ifthecorrelationis.30,move30%ofthedistancefromtheaveragetothematching
GPA.
Step1getsyouthebaseline,theGPAyouwouldhavepredictedifyouweretoldnothing
aboutJuliebeyondthefactthatsheisagraduatingsenior.Intheabsenceofinformation,
you would have predicted the average. (This is similar to assigning the base-rate
probability of business administration grahavрduates when you are told nothing about
Tom W.) Step 2 is your intuitive prediction, which matches your evaluation of the
evidence.Step3movesyoufromthebaselinetowardyourintuition,butthedistanceyou
areallowedtomovedependsonyourestimateofthecorrelation.Youendup,atstep4,
withapredictionthatisinfluencedbyyourintuitionbutisfarmoremoderate.
Thisapproachtopredictionisgeneral.Youcanapplyitwheneveryouneedtopredict
a quantitative variable, such as GPA, profit from an investment, or the growth of a
company.Theapproachbuildsonyourintuition,butitmoderatesit,regressesittoward
themean.Whenyouhavegoodreasonstotrusttheaccuracyofyourintuitiveprediction—
a strong correlation between the evidence and the prediction—the adjustment will be
small.
Intuitive predictions need to be corrected because they are not regressive and
thereforearebiased.SupposethatIpredictforeachgolferinatournamentthathisscore
on day 2 will be the same as his score on day 1. This prediction does not allow for
regressiontothemean:thegolferswhofaredwellonday1willonaveragedolesswell
on day 2, and those who did poorly will mostly improve. When they are eventually
comparedtoactualoutcomes,nonregressivepredictionswillbefoundtobebiased.They
are on average overly optimistic for those who did best on the first day and overly
pessimisticforthosewhohadabadstart.Thepredictionsareasextremeastheevidence.
Similarly, if you use childhood achievements to predict grades in college without
regressingyourpredictionstowardthemean,youwillmoreoftenthannotbedisappointed
bytheacademicoutcomesofearlyreadersandhappilysurprisedbythegradesofthose
who learned to read relatively late. The corrected intuitive predictions eliminate these
biases,sothatpredictions(bothhighandlow)areaboutequallylikelytooverestimateand
tounderestimatethetruevalue.Youstillmakeerrorswhenyourpredictionsareunbiased,
buttheerrorsaresmalleranddonotfavoreitherhighorlowoutcomes.
ADefenseofExtremePredictions?
IintroducedTomWearliertoillustratepredictionsofdiscreteoutcomessuchasfieldof
specialization or success in an examination, which are expressed by assigning a
probabilitytoaspecifiedevent(orinthatcasebyrankingoutcomesfromthemosttothe
leastprobable).Ialsodescribedaprocedurethatcountersthecommonbiasesofdiscrete
prediction:neglectofbaseratesandinsensitivitytothequalityofinformation.
Thebiaseswefindinpredictionsthatareexpressedonascale,suchasGPAorthe
revenue of a firm, are similar to the biases observed in judging the probabilities of
outcomes.
Thecorrectiveproceduresarealsosimilar:
Bothcontainabaselineprediction,whichyouwouldmakeifyouknewnothingabout
thecaseathand.Inthecategoricalcase,itwasthebaserate.Inthenumericalcase,it
istheaverageoutcomeintherelevantcategory.
Bothcontainanintuitiveprediction,whichexpressesthenumberthatcomestoyour
mind,whetheritisaprobabilityoraGPA.
Inbothcases,youaimforapredictionthatisintermediatebetweenthebaselineand
yourintuitiveresponse.
Inthedefaultcaseofnousefulevidence,youstaywiththebaseline.
Attheotherextreme,youalsostaywithyourinitialpredictiononsр.Thiswillhappen,
ofcourse,onlyifyouremaincompletelyconfidentinyourinitialpredictionaftera
criticalreviewoftheevidencethatsupportsit.
Inmostcasesyouwillfindsomereasontodoubtthatthecorrelationbetweenyour
intuitivejudgmentandthetruthisperfect,andyouwillendupsomewherebetween
thetwopoles.
Thisprocedureisanapproximationofthelikelyresultsofanappropriatestatistical
analysis. If successful, it will move you toward unbiased predictions, reasonable
assessments of probability, and moderate predictions of numerical outcomes. The two
procedures are intended to address the same bias: intuitive predictions tend to be
overconfidentandoverlyextreme.
CorrectingyourintuitivepredictionsisataskforSystem2.Significanteffortisrequiredto
find the relevant reference category, estimate the baseline prediction, and evaluate the
qualityoftheevidence.Theeffortisjustifiedonlywhenthestakesarehighandwhenyou
areparticularlykeennottomakemistakes.Furthermore,youshouldknowthatcorrecting
yourintuitionsmaycomplicateyourlife.Acharacteristicofunbiasedpredictionsisthat
they permit the prediction of rare or extreme events only when the information is very
good. If you expect your predictions to be of modest validity, you will never guess an
outcomethatiseitherrareorfarfromthemean.Ifyourpredictionsareunbiased,youwill
neverhavethesatisfyingexperienceofcorrectlycallinganextremecase.Youwillnever
beabletosay,“Ithoughtso!”whenyourbeststudentinlawschoolbecomesaSupreme
Courtjustice,orwhenastart-upthatyouthoughtverypromisingeventuallybecomesa
majorcommercialsuccess.Giventhelimitationsoftheevidence,youwillneverpredict
thatanoutstandinghighschoolstudentwillbeastraight-AstudentatPrinceton.Forthe
samereason,aventure capitalist will neverbetoldthatthe probabilityofsuccessfora
start-upinitsearlystagesis“veryhigh.”
The objections to the principle of moderating intuitive predictions must be taken
seriously, because absence of bias is not always what matters most. A preference for
unbiasedpredictionsisjustifiedifallerrorsofpredictionaretreatedalike,regardlessof
their direction. But there are situations in which one type of error is much worse than
another.Whenaventurecapitalistlooksfor“thenextbigthing,”theriskofmissingthe
next Google or Facebook is far more important than the risk of making a modest
investmentinastart-upthatultimatelyfails.Thegoalofventurecapitalistsistocallthe
extreme cases correctly, even at the cost of overestimating the prospects of many other
ventures. For a conservative banker making large loans, the risk of a single borrower
going bankrupt may outweigh the risk of turning down several would-be clients who
would fulfill their obligations. In such cases, the use of extreme language (“very good
prospect,” “serious risk of default”) may have some justification for the comfort it
provides,eveniftheinformationonwhichthesejudgmentsarebasedisofonlymodest
validity.
Forarationalperson,predictionsthatareunbiasedandmoderateshouldnotpresenta
problem.Afterall,therationalventurecapitalistknowsthateventhemostpromisingstart-
ups have only a moderate chance of success. She views her job as picking the most
promisingbetsfromthebetsthatareavailableanddoesnotfeeltheneedtodeludeherself
about the prospects of a start-up in which she plans to invest. Similarly, rational
individualspredictingtherevenueofafirmwillnotbeboundtoasingleysрnumber—
theyshouldconsidertherangeofuncertaintyaroundthemostlikelyoutcome.Arational
personwillinvestalargesuminanenterprisethatismostlikelytofailiftherewardsof
successarelargeenough,withoutdeludingherselfaboutthechancesofsuccess.However,
we are not all rational, and some of us may need the security of distorted estimates to
avoid paralysis. If you choose to delude yourself by accepting extreme predictions,
however,youwilldowelltoremainawareofyourself-indulgence.
PerhapsthemostvaluablecontributionofthecorrectiveproceduresIproposeisthat
theywillrequireyoutothinkabouthowmuchyouknow.Iwilluseanexamplethatis
familiarintheacademicworld,buttheanalogiestootherspheresoflifeareimmediate.A
department is about to hire a young professor and wants to choose the one whose
prospects for scientific productivity are the best. The search committee has narrowed
downthechoicetotwocandidates:
Kim recently completed her graduate work. Her recommendations are spectacular
andshegave a brillianttalkandimpressedeveryoneinherinterviews.Shehasno
substantialtrackrecordofscientificproductivity.
Jane has held a postdoctoral position for the last three years. She has been very
productiveandherresearchrecordisexcellent,buthertalkandinterviewswereless
sparklingthanKim’s.
TheintuitivechoicefavorsKim,becausesheleftastrongerimpression,andWYSIATI.
ButitisalsothecasethatthereismuchlessinformationaboutKimthanaboutJane.We
arebacktothelawofsmallnumbers.Ineffect,youhaveasmallersampleofinformation
fromKimthanfromJane,andextremeoutcomesaremuchmorelikelytobeobservedin
small samples. There is more luck in the outcomes of small samples, and you should
therefore regress your prediction more deeply toward the mean in your prediction of
Kim’sfutureperformance.WhenyouallowforthefactthatKimislikelytoregressmore
thanJane,youmightendupselectingJanealthoughyouwerelessimpressedbyher.In
the context of academic choices, I would vote for Jane, but it would be a struggle to
overcomemyintuitiveimpressionthatKimismorepromising.Followingourintuitionsis
morenatural,andsomehowmorepleasant,thanactingagainstthem.
You can readily imagine similar problems in different contexts, such as a venture
capitalistchoosingbetweeninvestmentsintwostart-upsthatoperateindifferentmarkets.
Onestart-uphas a productforwhich demandcanbeestimated with fairprecision.The
other candidate is more exciting and intuitively promising, but its prospects are less
certain.Whetherthebestguessabouttheprospectsofthesecondstart-upisstillsuperior
whentheuncertaintyisfactoredinisaquestionthatdeservescarefulconsideration.
ATwo-SystemsViewofRegression
Extremepredictionsandawillingnesstopredictrareeventsfromweakevidenceareboth
manifestations of System 1. It is natural for the associative machinery to match the
extremenessofpredictionstotheperceivedextremenessofevidenceonwhichitisbased
—thisishowsubstitutionworks.AnditisnaturalforSystem1togenerateoverconfident
judgments,becauseconfidence,aswehaveseen,isdeterminedbythecoherenceofthe
beststoryyoucantellfromtheevidenceathand.Bewarned:yourintuitionswilldeliver
predictionsthataretooextremeandyouwillbeinclineheрdtoputfartoomuchfaithin
them.
RegressionisalsoaproblemforSystem2.Theveryideaofregressiontothemeanis
alien and difficult to communicate and comprehend. Galton had a hard time before he
understoodit.Manystatisticsteachersdreadtheclassinwhichthetopiccomesup,and
theirstudentsoftenendupwithonlyavagueunderstandingofthiscrucialconcept.Thisis
acasewhereSystem2requiresspecialtraining.Matchingpredictionstotheevidenceis
notonlysomethingwedointuitively;italsoseemsareasonablethingtodo.Wewillnot
learntounderstandregressionfromexperience.Evenwhenaregressionisidentified,as
wesawinthestoryoftheflightinstructors,itwillbegivenacausalinterpretationthatis
almostalwayswrong.
SpeakingofIntuitivePredictions
“That start-up achieved an outstanding proof of concept, but we shouldn’t expect
themtodoaswellinthefuture.Theyarestillalongwayfromthemarketandthere
isalotofroomforregression.”
“Ourintuitivepredictionisveryfavorable,butitisprobablytoohigh.Let’stakeinto
accountthestrengthofourevidenceandregressthepredictiontowardthemean.”
“Theinvestmentmaybeagoodidea,evenifthebestguessisthatitwillfail.Let’s
notsaywereallybelieveitisthenextGoogle.”
“Ireadonereviewofthatbrandanditwasexcellent.Still,thatcouldhavebeena
fluke.Let’sconsideronlythebrandsthathavealargenumberofreviewsandpickthe
onethatlooksbest.”
P
Part3
P
Overconfidence
P
TheIllusionofUnderstanding
Thetrader-philosopher-statisticianNassimTalebcouldalsobeconsideredapsychologist.
InThe Black Swan, Taleb introduced the notion of a narrative fallacy to describe how
flawedstoriesofthepastshapeourviewsoftheworldandourexpectationsforthefuture.
Narrative fallacies arise inevitably from our continuous attempt to make sense of the
world.Theexplanatorystoriesthatpeoplefindcompellingaresimple;areconcreterather
thanabstract;assignalargerroletotalent,stupidity,andintentionsthantoluck;andfocus
onafewstrikingeventsthathappenedratherthanonthecountlesseventsthatfailedto
happen.Anyrecentsalienteventisacandidatetobecomethekernelofacausalnarrative.
Talebsuggeststhatwehumansconstantlyfoolourselvesbyconstructingflimsyaccounts
ofthepastandbelievingtheyaretrue.
Goodstoriesprovideasimpleandcoherentaccount>
Acompellingnarrativefostersanillusionofinevitability.Considerthestoryofhow
Googleturnedintoagiantofthetechnologyindustry.Twocreativegraduatestudentsin
thecomputersciencedepartmentatStanfordUniversitycomeupwithasuperiorwayof
searchinginformationontheInternet.Theyseekandobtainfundingtostartacompany
andmakeaseriesofdecisionsthatworkoutwell.Withinafewyears,thecompanythey
started is one of the most valuable stocks in America, and the two former graduate
students are among the richest people on the planet. On one memorable occasion, they
werelucky,whichmakesthestoryevenmorecompelling:ayearafterfoundingGoogle,
theywere willingto sell their company forless than$1 million, but the buyersaid the
price was too high. Mentioning the single lucky incident actually makes it easier to
underestimatethemultitudeofwaysinwhichluckaffectedtheoutcome.
A detailed history would specify the decisions of Google’s founders, but for our
purposes it suffices to say that almost every choice they made had a good outcome. A
morecompletenarrativewoulddescribetheactionsofthefirmsthatGoogledefeated.The
haplesscompetitorswouldappeartobeblind,slow,andaltogetherinadequateindealing
withthethreatthateventuallyoverwhelmedthem.
Iintentionallytoldthistaleblandly,butyougettheidea:thereisaverygoodstory
here.Fleshedoutinmoredetail,thestorycouldgiveyouthesensethatyouunderstand
whatmadeGooglesucceed;itwouldalsomakeyoufeelthatyouhavelearnedavaluable
generallessonaboutwhatmakesbusinessessucceed.Unfortunately,thereisgoodreason
tobelievethatyoursenseofunderstandingandlearningfromtheGooglestoryislargely
illusory. The ultimate test of an explanation is whether it would have made the event
predictableinadvance.NostoryofGoogle’sunlikelysuccesswillmeetthattest,because
nostorycanincludethemyriadofeventsthatwouldhavecausedadifferentoutcome.The
human mind does not deal well with nonevents. The fact that many of the important
eventsthatdidoccurinvolvechoicesfurthertemptsyoutoexaggeratetheroleofskilland
underestimate the part that luck played in the outcome. Because every critical decision
turnedoutwell,therecordsuggestsalmostflawlessprescience—butbadluckcouldhave
disruptedanyoneofthesuccessfulsteps.Thehaloeffectaddsthefinaltouches,lending
anauraofinvincibilitytotheheroesofthestory.
Likewatchingaskilledrafteravoidingonepotentialcalamityafteranotherashegoes
downtherapids,theunfoldingoftheGooglestoryisthrillingbecauseoftheconstantrisk
of disaster. However, there is foр an instructive difference between the two cases. The
skilledrafterhasgonedownrapidshundredsoftimes.Hehaslearnedtoreadtheroiling
water in front of him and to anticipate obstacles. He has learned to make the tiny
adjustments of posture that keep him upright. There are fewer opportunities for young
mentolearnhowtocreateagiantcompany,andfewerchancestoavoidhiddenrocks—
suchasabrilliantinnovationbyacompetingfirm.Ofcoursetherewasagreatdealofskill
intheGooglestory,butluckplayedamoreimportantroleintheactualeventthanitdoes
inthetellingofit.Andthemoreluckwasinvolved,thelessthereistobelearned.
At work here is that powerful WY SIATI rule. You cannot help dealing with the
limitedinformationyouhaveasifitwereallthereistoknow.Youbuildthebestpossible
story from the information available to you, and if it is a good story, you believe it.
Paradoxically,itiseasiertoconstructacoherentstorywhenyouknowlittle,whenthere
arefewerpieces to fitintothe puzzle. Ourcomfortingconvictionthat theworldmakes
senserestsonasecurefoundation:ouralmostunlimitedabilitytoignoreourignorance.
Ihaveheardoftoomanypeoplewho“knewwellbeforeithappenedthatthe2008
financialcrisiswasinevitable.”Thissentencecontainsahighlyobjectionableword,which
shouldberemovedfromourvocabularyindiscussionsofmajorevents.Thewordis,of
course,knew.Somepeoplethoughtwellinadvancethattherewouldbeacrisis,butthey
didnotknowit.Theynowsaytheyknewitbecausethecrisisdidinfacthappen.Thisisa
misuse of an important concept. In everyday language, we apply the word know only
whenwhatwasknownistrueandcanbeshowntobetrue.Wecanknowsomethingonly
ifitisbothtrueandknowable.Butthepeoplewhothoughttherewouldbeacrisis(and
therearefewerofthemthannowrememberthinkingit)couldnotconclusivelyshowitat
thetime.Manyintelligentandwell-informedpeoplewerekeenlyinterestedinthefuture
oftheeconomyanddidnotbelieveacatastrophewasimminent;Iinferfromthisfactthat
thecrisiswasnotknowable.Whatisperverseabouttheuseofknowinthiscontextisnot
that some individuals get credit for prescience that they do not deserve. It is that the
language implies that the world is more knowable than it is. It helps perpetuate a
perniciousillusion.
Thecoreoftheillusionisthatwebelieveweunderstandthepast,whichimpliesthat
thefuturealsoshouldbeknowable,butinfactweunderstandthepastlessthanwebelieve
wedo.Knowisnottheonlywordthatfostersthisillusion.Incommonusage,thewords
intuitionandpremonitionalsoarereservedforpastthoughtsthatturnedouttobetrue.The
statement“Ihadapremonitionthatthemarriagewouldnotlast,butIwaswrong”sounds
odd,asdoesanysentenceaboutanintuitionthatturnedouttobefalse.Tothinkclearly
aboutthefuture,weneedtocleanupthelanguagethatweuseinlabelingthebeliefswe
hadinthepast.
TheSocialCostsofHindsight
The mind that makes up narratives about the past is a sense-making organ. When an
unpredictedeventoccurs,weimmediatelyadjustourviewoftheworldtoaccommodate
the surprise. Imagine yourself before a football game between two teams that have the
samerecordofwinsandlosses.Nowthegameisover,andoneteamtrashedtheother.In
yourrevisedmodeloftheworld,thewinningteamismuchstrongerthantheloser,and
yourviewofthepastaswellasofthefuturehasbeenalteredbefрythatnewperception.
Learning from surprises is a reasonable thing to do, but it can have some dangerous
consequences.
A general limitation of the human mind is its imperfect ability to reconstruct past
states of knowledge, or beliefs that have changed. Once you adopt a new view of the
world(orofanypartofit),youimmediatelylosemuchofyourabilitytorecallwhatyou
usedtobelievebeforeyourmindchanged.
Many psychologists have studied what happens when people change their minds.
Choosingatopiconwhichmindsarenotcompletelymadeup—say,thedeathpenalty—
theexperimentercarefullymeasurespeople’sattitudes.Next,theparticipantsseeorheara
persuasiveproorconmessage.Thentheexperimentermeasurespeople’sattitudesagain;
they usually are closer to the persuasive message they were exposed to. Finally, the
participantsreporttheopiniontheyheldbeforehand.Thistaskturnsouttobesurprisingly
difficult. Asked to reconstruct their former beliefs, people retrieve their current ones
instead—an instance of substitution—and many cannot believe that they ever felt
differently.
Yourinabilitytoreconstruct pastbeliefswill inevitably causeyouto underestimate
the extent to which you were surprised by past events. Baruch Fischh off first
demonstratedthis“I-knew-it-all-along”effect,orhindsightbias,whenhewasastudentin
Jerusalem. Together with Ruth Beyth (another of our students), Fischh off conducted a
surveybeforePresidentRichardNixonvisitedChinaandRussiain1972.Therespondents
assigned probabilities to fifteen possible outcomes of Nixon’s diplomatic initiatives.
WouldMaoZedongagreetomeetwithNixon?MighttheUnitedStatesgrantdiplomatic
recognition to China? After decades of enmity, could the United States and the Soviet
Unionagreeonanythingsignificant?
AfterNixon’sreturnfromhistravels,FischhoffandBeythaskedthesamepeopleto
recall the probability that they had originally assigned to each of the fifteen possible
outcomes.Theresultswereclear.Ifaneventhadactuallyoccurred,peopleexaggerated
theprobabilitythattheyhadassignedtoitearlier.Ifthepossibleeventhadnotcometo
pass, the participants erroneously recalled that they had always considered it unlikely.
Furtherexperimentsshowedthatpeopleweredriventooverstatetheaccuracynotonlyof
theiroriginalpredictionsbutalsoofthosemadebyothers.Similarresultshavebeenfound
forothereventsthatgrippedpublicattention,suchastheO.J.Simpsonmurdertrialand
the impeachment of President Bill Clinton. The tendency to revise the history of one’s
beliefsinlightofwhatactuallyhappenedproducesarobustcognitiveillusion.
Hindsightbiashasperniciouseffectsontheevaluationsofdecisionmakers.Itleads
observerstoassessthequalityofadecisionnotbywhethertheprocesswassoundbutby
whetheritsoutcomewasgoodorbad.Consideralow-risksurgicalinterventioninwhich
anunpredictableaccidentoccurredthatcausedthepatient’sdeath.Thejurywillbeprone
to believe, after the fact, that the operation was actually risky and that the doctor who
ordered it should have known better. This outcome bias makes it almost impossible to
evaluate a decision properly—in terms of the beliefs that were reasonable when the
decisionwasmade.
Hindsight is especially unkind to decision makers who act as agents for others—
physicians, financial advisers, third-base coaches, CEOs, social workers, diplomats,
politicians. We are prone to blame decision makers for good decisions that worked out
badlyandtogivethemtoolittlecreditforsuccessfulmovesecaрthatappearobviousonly
afterthefact.Thereisaclearoutcomebias.Whentheoutcomesarebad,theclientsoften
blame their agents for not seeing the handwriting on the wall—forgetting that it was
writtenininvisibleinkthatbecamelegibleonlyafterward.Actionsthatseemedprudentin
foresight can look irresponsibly negligent in hindsight. Based on an actual legal case,
students in California were asked whether the city of Duluth, Minnesota, should have
shoulderedtheconsiderablecostofhiringafull-timebridgemonitortoprotectagainstthe
riskthatdebrismightgetcaughtandblockthefreeflowofwater.Onegroupwasshown
onlytheevidenceavailableatthetimeofthecity’sdecision;24%ofthesepeoplefeltthat
Duluth should take on the expense of hiring a flood monitor. The second group was
informed that debris had blocked the river, causing major flood damage; 56% of these
people said the city should have hired the monitor, although they had been explicitly
instructednottolethindsightdistorttheirjudgment.
The worse the consequence, the greater the hindsight bias. In the case of a
catastrophe,suchas9/11,weareespeciallyreadytobelievethattheofficialswhofailedto
anticipateitwerenegligentorblind.OnJuly10,2001,theCentralIntelligenceAgency
obtainedinformationthatal-QaedamightbeplanningamajorattackagainsttheUnited
States.GeorgeTenet,directoroftheCIA,broughttheinformationnottoPresidentGeorge
W. Bush but to National Security Adviser Condoleezza Rice. When the facts later
emerged,BenBradlee,thelegendaryexecutiveeditorofTheWashingtonPost,declared,
“Itseemstomeelementarythatifyou’vegotthestorythat’sgoingtodominatehistory
youmightaswellgorighttothepresident.”ButonJuly10,nooneknew—orcouldhave
known—thatthistidbitofintelligencewouldturnouttodominatehistory.
Because adherence to standard operating procedures is difficult to second-guess,
decisionmakerswhoexpecttohavetheirdecisionsscrutinizedwithhindsightaredriven
to bureaucratic solutions—and to an extreme reluctance to take risks. As malpractice
litigationbecamemorecommon,physicianschangedtheirproceduresinmultipleways:
ordered more tests, referred more cases to specialists, applied conventional treatments
evenwhentheywereunlikelytohelp.Theseactionsprotectedthephysiciansmorethan
they benefited the patients, creating the potential for conflicts of interest. Increased
accountabilityisamixedblessing.
Although hindsight and the outcome bias generally foster risk aversion, they also
bring undeserved rewards to irresponsible risk seekers, such as a general or an
entrepreneurwhotookacrazygambleandwon.Leaderswhohavebeenluckyarenever
punishedforhavingtakentoomuchrisk.Instead,theyarebelievedtohavehadtheflair
andforesighttoanticipatesuccess,andthesensiblepeoplewhodoubtedthemareseenin
hindsightasmediocre,timid,andweak.Afewluckygamblescancrownarecklessleader
withahaloofprescienceandboldness.
RecipesforSuccess
Thesense-makingmachineryofSystem1makesusseetheworldasmoretidy,simple,
predictable,andcoherentthan itreallyis.The illusionthatonehas understoodthepast
feedsthefurtherillusionthatonecanpredictandcontrolthefuture.Theseillusionsare
comforting.Theyreducetheanxietythatwewouldexperienceifweallowedourselvesto
fully acknowledge the uncertainties of existence. We all have a need for the reassuring
messagethatactionshaveappropriateconsequences,andthatsuccesswillrewardwisdom
andcourage.Manybdecрusinessbooksaretailor-madetosatisfythisneed.
Doleadersandmanagementpracticesinfluencetheoutcomesoffirmsinthemarket?
Of course they do, and the effects have been confirmed by systematic research that
objectivelyassessedthecharacteristicsofCEOsandtheirdecisions,andrelatedthemto
subsequent outcomes of the firm. In one study, the CEOs were characterized by the
strategy of the companies they had led before their current appointment, as well as by
management rules and procedures adopted after their appointment. CEOs do influence
performance, but the effects are much smaller than a reading of the business press
suggests.
Researchersmeasurethestrengthofrelationshipsbyacorrelationcoefficient,which
variesbetween0and1.Thecoefficientwasdefinedearlier(inrelationtoregressiontothe
mean) by the extent to which two measures are determined by shared factors. A very
generousestimateofthecorrelationbetweenthesuccessofthefirmandthequalityofits
CEOmightbeashighas.30,indicating30%overlap.Toappreciatethesignificanceof
thisnumber,considerthefollowingquestion:
Supposeyouconsidermanypairsoffirms.Thetwofirmsineachpairaregenerally
similar,buttheCEOofoneofthemisbetterthantheother.Howoftenwillyoufind
thatthefirmwiththestrongerCEOisthemoresuccessfulofthetwo?
In a well-ordered and predictable world, the correlation would be perfect (1), and the
strongerCEOwouldbefoundtoleadthemoresuccessfulfirmin100%ofthepairs.Ifthe
relativesuccessofsimilarfirmswasdeterminedentirelybyfactorsthattheCEOdoesnot
control(callthemluck,ifyouwish),youwouldfindthemoresuccessfulfirmledbythe
weaker CEO 50% of the time. A correlation of .30 implies that you would find the
strongerCEOleadingthestrongerfirminabout60%ofthepairs—animprovementofa
mere 10 percentage points over random guessing, hardly grist for the hero worship of
CEOswesooftenwitness.
Ifyouexpectedthisvaluetobehigher—andmostofusdo—thenyoushouldtakethat
asanindicationthatyouarepronetooverestimatethepredictabilityoftheworldyoulive
in.Makenomistake:improvingtheoddsofsuccessfrom1:1to3:2isaverysignificant
advantage,bothattheracetrackandinbusiness.Fromtheperspectiveofmostbusiness
writers, however, a CEO who has so little control over performance would not be
particularlyimpressiveevenifherfirmdidwell.Itisdifficulttoimaginepeopleliningup
atairportbookstorestobuyabookthatenthusiasticallydescribesthepracticesofbusiness
leaderswho,onaverage,dosomewhatbetterthanchance.Consumershaveahungerfora
clear message about the determinants of success and failure in business, and they need
storiesthatofferasenseofunderstanding,howeverillusory.
In his penetrating book The Halo Effect, Philip Rosenzweig, a business school
professorbasedinSwitzerland,showshowthedemandforillusorycertaintyismetintwo
populargenresofbusinesswriting:historiesoftherise(usually)andfall(occasionally)of
particularindividualsandcompanies,andanalysesofdifferencesbetweensuccessfuland
less successful firms. He concludes that stories of success and failure consistently
exaggerate the impact of leadership style and management practices on firm outcomes,
andthustheirmessageisrarelyuseful.
Toappreciatewhatisgoingon,imaginethatbusinessexperts,suchasotherCEOs,
areaskedtocommentonthereputationofthechiefexecutiveofacompany.Theypoрare
keenlyawareof whether thecompanyhas recently beenthrivingor failing. Aswesaw
earlierinthecaseofGoogle,thisknowledgegeneratesahalo.TheCEOofasuccessful
companyislikelytobecalledflexible,methodical,anddecisive.Imaginethatayearhas
passedandthingshavegonesour.Thesameexecutiveisnowdescribedasconfused,rigid,
andauthoritarian.Bothdescriptionssoundrightatthetime:itseemsalmostabsurdtocall
asuccessfulleaderrigidandconfused,orastrugglingleaderflexibleandmethodical.
Indeed, the halo effect is so powerful that you probably find yourself resisting the
idea that the same person and the same behaviors appear methodical when things are
goingwellandrigidwhenthingsaregoingpoorly.Becauseofthehaloeffect,wegetthe
causalrelationshipbackward:wearepronetobelievethatthefirmfailsbecauseitsCEO
isrigid,whenthetruthisthattheCEOappearstoberigidbecausethefirmisfailing.This
ishowillusionsofunderstandingareborn.
The halo effect and outcome bias combine to explain the extraordinary appeal of
books that seek to draw operational morals from systematic examination of successful
businesses. One of the best-known examples of this genre is Jim Collins and Jerry I.
Porras’s Built to Last. The book contains a thorough analysis of eighteen pairs of
competingcompanies,inwhichonewasmoresuccessfulthantheother.Thedataforthese
comparisonsareratingsofvariousaspectsofcorporateculture,strategy,andmanagement
practices. “We believe every CEO, manager, and entrepreneur in the world should read
thisbook,”theauthorsproclaim.“Youcanbuildavisionarycompany.”
ThebasicmessageofBuilttoLastandothersimilarbooksisthatgoodmanagerial
practicescanbeidentifiedandthatgoodpracticeswillberewardedbygoodresults.Both
messagesareoverstated.Thecomparisonoffirmsthathavebeenmoreorlesssuccessful
istoasignificantextentacomparisonbetweenfirmsthathavebeenmoreorlesslucky.
Knowing the importance of luck, you should be particularly suspicious when highly
consistentpatternsemergefromthecomparisonofsuccessfulandlesssuccessfulfirms.In
thepresenceofrandomness,regularpatternscanonlybemirages.
Becauseluckplaysalargerole,thequalityofleadershipandmanagementpractices
cannot be inferred reliably from observations of success. And even if you had perfect
foreknowledge that a CEO has brilliant vision and extraordinary competence, you still
wouldbeunabletopredicthowthecompanywillperformwithmuchbetteraccuracythan
theflipofacoin.Onaverage,thegapincorporateprofitabilityandstockreturnsbetween
theoutstandingfirmsandthelesssuccessfulfirmsstudiedinBuilttoLastshranktoalmost
nothing in the period following the study. The average profitability of the companies
identifiedinthefamousIn Search of Excellencedroppedsharplyas wellwithinashort
time. A study of Fortune’s “Most Admired Companies” finds that over a twenty-year
period,thefirmswiththeworstratingswentontoearnmuchhigherstockreturnsthanthe
mostadmiredfirms.
You are probably tempted to think of causal explanations for these observations:
perhaps the successful firms became complacent, the less successful firms tried harder.
Butthisisthewrongwaytothinkaboutwhathappened.Theaveragegapmustshrink,
because the original gap was due in good part to luck, which contributed both to the
success of the top firms and to the lagging performance of the rest. We have already
encounteredthisstatisticalfactoflife:regressiontothemean.
Storiesofhowbusinessesriseandfallstrikeachordwithreadersbyofferingwhatthe
humanmindneeds:asimplemessageoftriumphandfailurethatidentifiesclearcauses
and ignores the determinative power of luck and the inevitability of regression. These
stories induce and maintain an illusion of understanding, imparting lessons of little
enduringvaluetoreaderswhoarealltooeagertobelievethem.
SpeakingofHindsight
“Themistakeappearsobvious,butitisjusthindsight.Youcouldnothaveknownin
advance.”
“He’slearningtoomuchfromthissuccessstory,whichistootidy.Hehasfallenfora
narrativefallacy.”
“Shehasnoevidenceforsayingthatthefirmisbadlymanaged.Allsheknowsisthat
itsstockhasgonedown.Thisisanoutcomebias,parthindsightandparthaloeffect.”
“Let’snotfallfortheoutcomebias.Thiswasastupiddecisioneventhoughitworked
outwell.”
P
TheIllusionofValidity
System1isdesignedtojumptoconclusionsfromlittleevidence—anditisnotdesignedto
know the size of its jumps. Because of WYSIATI, only the evidence at hand counts.
Becauseofconfidencebycoherence,thesubjectiveconfidencewehaveinouropinions
reflects the coherence of the story that System 1 and System 2 have constructed. The
amount of evidence and its quality do not count for much, because poor evidence can
makeaverygoodstory.Forsomeofourmostimportantbeliefswehavenoevidenceat
all, except that people we love and trust hold these beliefs. Considering how little we
know,theconfidencewehaveinourbeliefsispreposterous—anditisalsoessential.
TheIllusionofValidity
ManydecadesagoIspentwhatseemedlikeagreatdealoftimeunderascorchingsun,
watching groups of sweaty soldiers as they solved a problem. I was doing my national
service in the Israeli Army at the time. I had completed an undergraduate degree in
psychology,andafterayearasaninfantryofficerwasassignedtothearmy’sPsychology
Branch, where one of my occasional duties was to help evaluate candidates for officer
training.WeusedmethodsthathadbeendevelopedbytheBritishArmyinWorldWarII.
Onetest,calledthe“leaderlessgroupchallenge,”wasconductedonanobstaclefield.
Eight candidates, strangers to each other, with all insignia of rank removed and only
numberedtagstoidentifythem,wereinstructedtoliftalonglogfromthegroundandhaul
ittoawallaboutsixfeethigh.Theentiregrouphadtogettotheothersideofthewall
withoutthelogtouchingeitherthegroundorthewall,andwithoutanyonetouchingthe
wall.Ifanyofthesethingshappened,theyhadtodeclareitsigрЉTandstartagain.
Therewasmorethanonewaytosolvetheproblem.Acommonsolutionwasforthe
teamtosendseveralmentotheothersidebycrawlingoverthepoleasitwasheldatan
angle, like a giant fishing rod, by other members of the group. Or else some soldiers
wouldclimbontosomeone’sshouldersandjumpacross.Thelastmanwouldthenhaveto
jumpupatthepole,heldupatananglebytherestofthegroup,shinnyhiswayalongits
lengthastheotherskepthimandthepolesuspendedintheair,andleapsafelytotheother
side.Failurewascommonatthispoint,whichrequiredthemtostartalloveragain.
AsacolleagueandImonitoredtheexercise,wemadenoteofwhotookcharge,who
tried to lead but was rebuffed, how cooperative each soldier was in contributing to the
group effort. We saw who seemed to be stubborn, submissive, arrogant, patient, hot-
tempered, persistent, or a quitter. We sometimes saw competitive spite when someone
whose idea had been rejected by the group no longer worked very hard. And we saw
reactionstocrisis:whoberatedacomradewhosemistakehadcausedthewholegroupto
fail,whosteppedforwardtoleadwhentheexhaustedteamhadtostartover.Underthe
stressoftheevent,wefelt,eachman’struenaturerevealeditself.Ourimpressionofeach
candidate’scharacterwasasdirectandcompellingasthecolorofthesky.
After watching the candidates make several attempts, we had to summarize our
impressionsofsoldiers’ leadershipabilitiesand determine, withanumerical score,who
should be eligible for officer training. We spent some time discussing each case and
reviewingourimpressions.Thetaskwasnotdifficult,becausewefeltwehadalreadyseen
eachsoldier’s leadershipskills. Some ofthe men hadlooked like strongleaders, others
hadseemedlikewimpsorarrogantfools,othersmediocrebutnothopeless.Quiteafew
lookedsoweakthatweruledthemoutascandidatesforofficerrank.Whenourmultiple
observations of each candidate converged on a coherent story, we were completely
confidentinourevaluationsandfeltthatwhatwehadseenpointeddirectlytothefuture.
Thesoldierwhotookoverwhenthegroupwasintroubleandledtheteamoverthewall
wasaleaderatthatmoment.Theobviousbestguessabouthowhewoulddointraining,or
incombat,wasthathewouldbeaseffectivethenashehadbeenatthewall.Anyother
predictionseemedinconsistentwiththeevidencebeforeoureyes.
Because our impressions of how well each soldier had performed were generally
coherent and clear, our formal predictions were just as definite. A single score usually
came to mind and we rarely experienced doubts or formed conflicting impressions. We
werequitewillingtodeclare,“Thisonewillnevermakeit,”“Thatfellowismediocre,but
he should do okay,” or “He will be a star.” We felt no need to question our forecasts,
moderatethem,orequivocate.Ifchallenged,however,wewerepreparedtoadmit,“Butof
courseanythingcouldhappen.”Wewerewillingtomakethatadmissionbecause,despite
our definite impressions about individual candidates, we knew with certainty that our
forecastswerelargelyuseless.
Theevidencethatwecouldnotforecastsuccessaccuratelywasoverwhelming.Every
fewmonthswehadafeedbacksessioninwhichwelearnedhowthecadetsweredoingat
the officer-training school and could compare our assessments against the opinions of
commanders who had been monitoring them for some time. The story was always the
same:ourabilitytopredictperformanceattheschoolwasnegligible.Ourforecastswere
betterthanblindguesses,butnotbymuch.
We weedredowncastfora while after receivingthediscouragingnews.But this
wasthearmy.Usefulornot,therewasaroutinetobefollowedandorderstobeobeyed.
Anotherbatchofcandidatesarrivedthenextday.Wetookthemtotheobstaclefield,we
facedthemwiththewall,theyliftedthelog,andwithinafewminuteswesawtheirtrue
naturesrevealed,asclearlyasbefore.Thedismaltruthaboutthequalityofourpredictions
had no effect whatsoever on how we evaluated candidates and very little effect on the
confidencewefeltinourjudgmentsandpredictionsaboutindividuals.
Whathappenedwasremarkable.Theglobalevidenceofourpreviousfailureshould
haveshakenourconfidenceinourjudgmentsofthecandidates,butitdidnot.Itshould
alsohavecausedustomoderateourpredictions,butitdidnot.Weknewasageneralfact
thatourpredictionswerelittlebetterthanrandomguesses,butwecontinuedtofeeland
act as if each of our specific predictions was valid. I was reminded of the Müller-Lyer
illusion, in which we know the lines are of equal length yet still see them as being
different. I was so struck by the analogy that I coined a term for our experience: the
illusionofvalidity.
Ihaddiscoveredmyfirstcognitiveillusion.
Decadeslater,Icanseemanyofthecentralthemesofmythinking—andofthisbook—in
thatoldstory.Ourexpectationsforthesoldiers’futureperformancewereaclearinstance
ofsubstitution,andoftherepresentativenessheuristicinparticular.Havingobservedone
hourofasoldier’sbehaviorinanartificialsituation,wefeltweknewhowwellhewould
facethechallengesofofficertrainingandofleadershipincombat.Ourpredictionswere
completelynonregressive—wehadnoreservationsaboutpredictingfailureoroutstanding
successfromweakevidence.ThiswasaclearinstanceofWYSIATI.Wehadcompelling
impressionsofthebehaviorweobservedandnogoodwaytorepresentourignoranceof
thefactorsthatwouldeventuallydeterminehowwellthecandidatewouldperformasan
officer.
Lookingback,themoststrikingpartofthestoryisthatourknowledgeofthegeneral
rule—thatwecouldnotpredict—hadnoeffectonourconfidenceinindividualcases.Ican
seenowthatourreactionwassimilartothatofNisbettandBorgida’sstudentswhenthey
were told that most people did not help a stranger suffering a seizure. They certainly
believedthestatisticstheywereshown,butthebaseratesdidnotinfluencetheirjudgment
ofwhetheranindividualtheysawonthevideowouldorwouldnothelpastranger.Justas
Nisbett and Borgida showed, people are often reluctant to infer the particular from the
general.
Subjectiveconfidencein ajudgmentisnotareasonedevaluationoftheprobability
thatthisjudgmentiscorrect.Confidenceisafeeling,whichreflectsthecoherenceofthe
information and the cognitive ease of processing it. It is wise to take admissions of
uncertainty seriously, but declarations of high confidence mainly tell you that an
individualhas constructedacoherentstoryinhismind,notnecessarilythatthestoryis
true.
TheIllusionofStock-PickingSkill
In1984,AmosandIandourfriendRichardThalervisitedaWallStreetfirm.Ourhost,a
senior investment manager, had invited us to discuss the role of judgment biases in
investing.IknewsolittleaboutfinancethatIdidnotevenknowwhattoaskhim,butI
remember one exchange. “When you sell a stock,” d nI asked, “who buys it?” He
answeredwithawaveinthevaguedirectionofthewindow,indicatingthatheexpected
thebuyertobesomeoneelseverymuchlikehim.Thatwasodd:Whatmadeoneperson
buyandtheothersell?Whatdidthesellersthinktheyknewthatthebuyersdidnot?
Sincethen,myquestionsaboutthestockmarkethavehardenedintoalargerpuzzle:a
major industry appears to be built largely on an illusion ofskill. Billions of shares are
tradedeveryday,withmanypeoplebuyingeachstockandotherssellingittothem.Itis
notunusualformorethan100millionsharesofasinglestocktochangehandsinoneday.
Mostofthebuyersandsellersknowthattheyhavethesameinformation;theyexchange
thestocksprimarilybecausetheyhavedifferentopinions.Thebuyersthinkthepriceistoo
low and likely to rise, while the sellers think the price is high and likely to drop. The
puzzleiswhybuyersandsellersalikethinkthatthecurrentpriceiswrong.Whatmakes
thembelievetheyknowmoreaboutwhatthepriceshouldbethanthemarketdoes?For
mostofthem,thatbeliefisanillusion.
Initsbroadoutlines,thestandardtheoryofhowthestockmarketworksisaccepted
by all the participants in the industry. Everybody in the investment business has read
Burton Malkiel’s wonderful book A Random Walk Down Wall Street. Malkiel’s central
ideaisthatastock’spriceincorporatesalltheavailableknowledgeaboutthevalueofthe
companyandthebestpredictionsaboutthefutureofthestock.Ifsomepeoplebelievethat
thepriceofastockwillbehighertomorrow,theywillbuymoreofittoday.This,inturn,
willcauseitspricetorise.Ifallassetsinamarketarecorrectlypriced,noonecanexpect
eithertogainortolosebytrading.Perfectpricesleavenoscopeforcleverness,butthey
alsoprotectfoolsfromtheirownfolly.Wenowknow,however,thatthetheoryisnotquite
right.Manyindividualinvestorsloseconsistentlybytrading,anachievementthatadart-
throwingchimpcouldnotmatch.Thefirstdemonstrationofthisstartlingconclusionwas
collectedbyTerryOdean,afinanceprofessoratUCBerkeleywhowasoncemystudent.
Odean began by studying the trading records of 10,000 brokerage accounts of
individual investors spanning a seven-year period. He was able to analyze every
transactiontheinvestorsexecutedthroughthatfirm,nearly163,000trades.Thisrichsetof
dataallowedOdeantoidentifyallinstancesinwhichaninvestorsoldsomeofhisholdings
in one stock and soon afterward bought another stock. By these actions the investor
revealedthathe(mostoftheinvestorsweremen)hadadefiniteideaaboutthefutureof
thetwostocks:heexpectedthestockthathechosetobuytodobetterthanthestockhe
chosetosell.
Todeterminewhetherthoseideaswerewellfounded,Odeancomparedthereturnsof
thestocktheinvestorhadsoldandthestockhehadboughtinitsplace,overthecourseof
oneyearafterthetransaction.Theresultswereunequivocallybad.Onaverage,theshares
thatindividualtraderssolddidbetterthanthosetheybought,byaverysubstantialmargin:
3.2 percentage points per year, above and beyond the significant costs of executing the
twotrades.
Itisimportanttorememberthatthisisastatementaboutaverages:someindividuals
didmuchbetter,othersdidmuchworse.However,itisclearthatforthelargemajorityof
individualinvestors,takingashoweranddoingnothingwouldhavebeenabetterpolicy
thanimplementingtheideas that cametotheir minds. Laterresearchby Odean andhis
colleagueBradBarbersupportedthisconclusion.Inapapertitled“TradingIsHazardous
toYourt-tWealth,”theyshowedthat,onaverage,themostactivetradershadthepoorest
results, while the investors who traded the least earned the highest returns. In another
paper, titled “Boys Will Be Boys,” they showed that men acted on their useless ideas
significantly more often than women, and that as a result women achieved better
investmentresultsthanmen.
Ofcourse,thereisalwayssomeoneontheothersideofeachtransaction;ingeneral,
thesearefinancialinstitutionsandprofessionalinvestors,whoarereadytotakeadvantage
ofthemistakesthatindividualtradersmakeinchoosingastocktosellandanotherstock
tobuy.FurtherresearchbyBarberandOdeanhasshedlightonthesemistakes.Individual
investorsliketolockintheirgainsbyselling“winners,”stocksthathaveappreciatedsince
they were purchased, and they hang on to their losers. Unfortunately for them, recent
winnerstendtodobetterthanrecentlosersintheshortrun,soindividualssellthewrong
stocks. They also buy the wrong stocks. Individual investors predictably flock to
companiesthatdrawtheirattentionbecausetheyareinthenews.Professionalinvestors
aremoreselectiveinrespondingtonews.Thesefindingsprovidesomejustificationforthe
labelof“smartmoney”thatfinanceprofessionalsapplytothemselves.
Although professionals are able to extract a considerable amount of wealth from
amateurs,fewstockpickers,ifany,havetheskillneededtobeatthemarketconsistently,
yearafteryear.Professionalinvestors,includingfundmanagers,failabasictestofskill:
persistentachievement.Thediagnosticfortheexistenceofanyskillistheconsistencyof
individualdifferencesinachievement.Thelogicissimple:ifindividualdifferencesinany
oneyeararedueentirelytoluck,therankingofinvestorsandfundswillvaryerratically
andtheyear-to-yearcorrelationwillbezero.Wherethereisskill,however,therankings
willbemorestable.Thepersistenceofindividualdifferencesisthemeasurebywhichwe
confirmtheexistenceofskillamonggolfers,carsalespeople,orthodontists,orspeedytoll
collectorsontheturnpike.
Mutualfundsarerunbyhighlyexperiencedandhardworkingprofessionalswhobuy
and sell stocks to achieve the best possible results for their clients. Nevertheless, the
evidencefrommorethanfiftyyearsofresearchisconclusive:foralargemajorityoffund
managers, the selection of stocks is more like rolling dice than like playing poker.
Typicallyatleasttwooutofeverythreemutualfundsunderperformtheoverallmarketin
anygivenyear.
Moreimportant,theyear-to-yearcorrelationbetweentheoutcomesofmutualfundsis
very small, barely higher than zero. The successful funds in any given year are mostly
lucky;they haveagoodrollofthedice.Thereisgeneralagreementamongresearchers
that nearly all stock pickers, whether they know it or not—and few of them do—are
playinga gameof chance. Thesubjective experienceof traders isthat theyare making
sensibleeducatedguessesinasituationofgreatuncertainty.Inhighlyefficientmarkets,
however,educatedguessesarenomoreaccuratethanblindguesses.
SomeyearsagoIhadanunusualopportunitytoexaminetheillusionoffinancialskillup
close.Ihadbeeninvitedtospeaktoagroupofinvestmentadvisersinafirmthatprovided
financial advice and other services to very wealthy clients. I asked for some data to
preparemypresentationandwasgrantedasmalltreasure:aspreadsheetsummarizingthe
investmentoutcomesofsometwenty-fiveanonymouswealthadvisers,foreachofeight
consecutive years. Each adviser’s scoof re for each year was his (most of them were
men)maindeterminantofhisyear-endbonus.Itwasasimplemattertoranktheadvisers
by their performance in each year and to determine whether there were persistent
differences in skill among them and whether the same advisers consistently achieved
betterreturnsfortheirclientsyearafteryear.
Toanswerthequestion,Icomputedcorrelationcoefficientsbetweentherankingsin
eachpairofyears:year1withyear2,year1withyear3,andsoonupthroughyear7with
year 8. That yielded 28 correlation coefficients, one for each pair of years. I knew the
theory and was prepared to find weak evidence of persistence of skill. Still, I was
surprisedtofindthattheaverageofthe28correlationswas.01.Inotherwords,zero.The
consistentcorrelationsthatwouldindicatedifferencesinskillwerenottobefound.The
resultsresembledwhatyouwouldexpectfromadice-rollingcontest,notagameofskill.
Nooneinthefirmseemedtobeawareofthenatureofthegamethatitsstockpickers
were playing. The advisers themselves felt they were competent professionals doing a
seriousjob,andtheirsuperiorsagreed.Ontheeveningbeforetheseminar,RichardThaler
andIhaddinnerwithsomeofthetopexecutivesofthefirm,thepeoplewhodecideonthe
sizeofbonuses. We askedthemtoguess theyear-to-yearcorrelationin the rankingsof
individual advisers. They thought they knew what was coming and smiled as they said
“notveryhigh”or“performancecertainlyfluctuates.”Itquicklybecameclear,however,
thatnooneexpectedtheaveragecorrelationtobezero.
Ourmessagetotheexecutiveswasthat,atleastwhenitcametobuildingportfolios,
thefirmwasrewardingluckasifitwereskill.Thisshouldhavebeenshockingnewsto
them,butitwasnot.Therewasnosignthattheydisbelievedus.Howcouldthey?After
all, we had analyzed their own results, and they were sophisticated enough to see the
implications,whichwepolitelyrefrainedfromspellingout.Weallwentoncalmlywith
ourdinner,andIhavenodoubtthatbothourfindingsandtheirimplicationswerequickly
sweptundertherugandthatlifeinthefirmwentonjustasbefore.Theillusionofskillis
notonlyanindividualaberration;itisdeeplyingrainedinthecultureoftheindustry.Facts
thatchallengesuchbasicassumptions—andtherebythreatenpeople’slivelihoodandself-
esteem—aresimplynotabsorbed.Theminddoesnotdigestthem.Thisisparticularlytrue
of statistical studies of performance, which provide base-rate information that people
generallyignorewhenitclasheswiththeirpersonalimpressionsfromexperience.
Thenextmorning,we reported the findingstotheadvisers,andtheirresponsewas
equallybland.Theirownexperienceofexercisingcarefuljudgmentoncomplexproblems
wasfarmorecompellingtothemthananobscurestatisticalfact.Whenweweredone,one
oftheexecutivesIhaddinedwiththepreviouseveningdrovemetotheairport.Hetold
me,withatraceofdefensiveness,“Ihavedoneverywellforthefirmandnoonecantake
thatawayfromme.”Ismiledandsaidnothing.ButIthought,“Well,Itookitawayfrom
you this morning. If your success was due mostly to chance, how much credit are you
entitledtotakeforit?”
WhatSupportstheIllusionsofSkillandValidity?
Cognitiveillusionscanbemorestubbornthanvisualillusions.Whatyoulearnedaboutthe
Müller-Lyer illusion did not change the way you see the lines, but it changed your
behavior.Younowknowthatyoucannottrustyourimpressionofthelengllithoflines
that have fins appended to them, and you also know that in the standard Müller-Lyer
displayyoucannottrustwhatyousee.Whenaskedaboutthelengthofthelines,youwill
reportyourinformedbelief,nottheillusionthatyoucontinuetosee.Incontrast,whenmy
colleaguesandIinthearmylearnedthatourleadershipassessmenttestshadlowvalidity,
we accepted that fact intellectually, but it had no impact on either our feelings or our
subsequent actions. The response we encountered in the financial firm was even more
extreme. I am convinced that the message that Thaler and I delivered to both the
executivesandtheportfoliomanagerswasinstantlyputawayinadarkcornerofmemory
whereitwouldcausenodamage.
Whydoinvestors,bothamateurandprofessional,stubbornlybelievethattheycando
better than the market, contrary to an economic theory that most of them accept, and
contrary to what they could learn from a dispassionate evaluation of their personal
experience?Manyofthethemesofpreviouschapterscomeupagainintheexplanationof
theprevalenceandpersistenceofanillusionofskillinthefinancialworld.
Themostpotentpsychologicalcauseoftheillusioniscertainlythatthepeoplewho
pick stocks are exercising high-level skills. They consult economic data and forecasts,
they examine income statements and balance sheets, they evaluate the quality of top
management, and they assess the competition. All this is serious work that requires
extensivetraining,andthepeoplewhodoithavetheimmediate(andvalid)experienceof
usingtheseskills.Unfortunately,skillinevaluatingthebusinessprospectsofafirmisnot
sufficientforsuccessfulstocktrading,wherethekeyquestioniswhethertheinformation
aboutthefirmisalreadyincorporatedinthepriceofitsstock.Tradersapparentlylackthe
skilltoanswerthiscrucialquestion,buttheyappeartobeignorantoftheirignorance.AsI
had discovered from watching cadets on the obstacle field, subjective confidence of
tradersisafeeling,notajudgment.Ourunderstandingofcognitiveeaseandassociative
coherencelocatessubjectiveconfidencefirmlyinSystem1.
Finally, the illusions of validity and skill are supported by a powerful professional
culture. We know that people can maintain an unshakable faith in any proposition,
howeverabsurd,whentheyaresustainedbyacommunityoflike-mindedbelievers.Given
theprofessionalcultureofthefinancialcommunity,itisnotsurprisingthatlargenumbers
ofindividualsinthatworldbelievethemselvestobeamongthechosenfewwhocando
whattheybelieveotherscannot.
TheIllusionsofPundits
Theideathatthefutureisunpredictableisunderminedeverydaybytheeasewithwhich
thepastisexplained.AsNassimTalebpointedoutinTheBlackSwan,ourtendencyto
constructandbelievecoherentnarrativesofthepastmakesitdifficultforustoacceptthe
limitsofourforecastingability.Everythingmakessenseinhindsight,afactthatfinancial
punditsexploiteveryeveningastheyofferconvincingaccountsoftheday’sevents.And
wecannotsuppressthepowerfulintuitionthatwhatmakessenseinhindsighttodaywas
predictableyesterday.Theillusionthatweunderstandthepastfostersoverconfidencein
ourabilitytopredictthefuture.
Theoften-usedimageofthe“marchofhistory”impliesorderanddirection.Marches,
unlikestrollsorwalks,arenotrandom.Wethinkthatweshouldbeabletoexplainthepast
byfocusingoneitherlargesocialmovementsandculturalandtechnologicaldevelopments
ortheintentionsandabilitiesofafewgcoreatmen.Theideathatlargehistoricalevents
aredeterminedbyluckisprofoundlyshocking,althoughitisdemonstrablytrue.Itishard
to think of the history of the twentieth century, including its large social movements,
withoutbringingintheroleofHitler,Stalin,andMaoZedong.Buttherewasamomentin
time,justbeforeaneggwasfertilized,whentherewasafifty-fiftychancethattheembryo
thatbecameHitlercouldhavebeenafemale.Compoundingthethreeevents,therewasa
probabilityofone-eighthofatwentiethcenturywithoutanyofthethreegreatvillainsand
itisimpossibletoarguethathistorywouldhavebeenroughlythesameintheirabsence.
Thefertilizationofthesethreeeggshadmomentousconsequences,anditmakesajokeof
theideathatlong-termdevelopmentsarepredictable.
Yettheillusionofvalidpredictionremainsintact,afactthatisexploitedbypeople
whose business is prediction—not only financial experts but pundits in business and
politics, too. Television and radio stations and newspapers have their panels of experts
whosejobitistocommentontherecentpastandforetellthefuture.Viewersandreaders
havetheimpressionthattheyarereceivinginformationthatissomehowprivileged,orat
least extremely insightful. And there is no doubt that the pundits and their promoters
genuinelybelievetheyareofferingsuchinformation.PhilipTetlock,apsychologistatthe
University of Pennsylvania, explained these so-called expert predictions in a landmark
twenty-yearstudy,whichhepublishedinhis2005bookExpertPoliticalJudgment:How
GoodIsIt?HowCanWeKnow?Tetlockhassetthetermsforanyfuturediscussionofthis
topic.
Tetlock interviewed 284 people who made their living “commenting or offering
adviceonpoliticalandeconomictrends.”Heaskedthemtoassesstheprobabilitiesthat
certaineventswouldoccurinthenottoodistantfuture,bothinareasoftheworldinwhich
theyspecializedandinregionsaboutwhichtheyhadlessknowledge.WouldGorbachev
be ousted in a coup? Would the United States go to war in the Persian Gulf? Which
countrywouldbecomethenextbigemergingmarket?Inall,Tetlockgatheredmorethan
80,000 predictions. He also asked the experts how they reached their conclusions, how
theyreactedwhenprovedwrong,andhowtheyevaluatedevidencethatdidnotsupport
their positions. Respondents were asked to rate the probabilities of three alternative
outcomes in every case: the persistence of the status quo, more of something such as
politicalfreedomoreconomicgrowth,orlessofthatthing.
Theresultsweredevastating.Theexpertsperformedworsethantheywouldhaveif
theyhadsimplyassigned equal probabilitiestoeachofthethreepotentialoutcomes.In
otherwords,peoplewhospendtheirtime,andearntheirliving,studyingaparticulartopic
producepoorerpredictionsthandart-throwingmonkeyswhowouldhavedistributedtheir
choices evenly over the options. Even in the region they knew best, experts were not
significantlybetterthannonspecialists.
Thosewho knowmore forecastvery slightlybetter thanthose who know less. But
thosewiththemostknowledgeareoftenlessreliable.Thereasonisthatthepersonwho
acquires more knowledge develops an enhanced illusion of her skill and becomes
unrealistically overconfident. “We reach the point of diminishing marginal predictive
returnsforknowledgedisconcertinglyquickly,”Tetlockwrites.“Inthisageofacademic
hyperspecialization, there is no reason for supposing that contributors to top journals—
distinguished political scientists, area study specialists, economists, and so on—are any
betterthanjournalistsorattentivereadersofTheNewYorkTimesin‘reading&#oul8217;
emerging situations.” The more famous the forecaster, Tetlock discovered, the more
flamboyanttheforecasts.“Expertsindemand,”hewrites,“weremoreoverconfidentthan
theircolleagueswhoekedoutexistencesfarfromthelimelight.”
Tetlockalsofoundthatexpertsresistedadmittingthattheyhadbeenwrong,andwhen
theywerecompelledtoadmiterror,theyhadalargecollectionofexcuses:theyhadbeen
wrongonlyintheirtiming,anunforeseeableeventhadintervened,ortheyhadbeenwrong
butfortherightreasons.Expertsarejusthumanintheend.Theyaredazzledbytheirown
brillianceandhatetobewrong.Expertsareledastraynotbywhattheybelieve,butby
how they think, says Tetlock. He uses the terminology from Isaiah Berlin’s essay on
Tolstoy, “The Hedgehog and the Fox.” Hedgehogs “know one big thing” and have a
theoryabouttheworld;theyaccountforparticulareventswithinacoherentframework,
bristlewithimpatiencetowardthosewhodon’tseethingstheirway,andareconfidentin
theirforecasts.Theyarealsoespeciallyreluctanttoadmiterror.Forhedgehogs,afailed
prediction is almost always “off only on timing” or “very nearly right.” They are
opinionatedandclear,whichisexactlywhattelevisionproducerslovetoseeonprograms.
Two hedgehogs on different sides of an issue, each attacking the idiotic ideas of the
adversary,makeforagoodshow.
Foxes,bycontrast,arecomplexthinkers.Theydon’tbelievethatonebigthingdrives
themarchofhistory(forexample,theyareunlikelytoaccepttheviewthatRonaldReagan
single-handedlyendedthecoldwarbystandingtallagainsttheSovietUnion).Insteadthe
foxes recognize that reality emerges from the interactions of many different agents and
forces,includingblindluck,oftenproducinglargeandunpredictableoutcomes.Itwasthe
foxeswhoscoredbestinTetlock’sstudy,althoughtheirperformancewasstillverypoor.
Theyarelesslikelythanhedgehogstobeinvitedtoparticipateintelevisiondebates.
ItisNottheExperts’Fault—TheWorldisDifficult
Themainpointofthischapterisnotthatpeoplewhoattempttopredictthefuturemake
many errors; that goes without saying. The first lesson is that errors of prediction are
inevitable because the world is unpredictable. The second is that high subjective
confidenceisnottobetrustedasanindicatorofaccuracy(lowconfidencecouldbemore
informative).
Short-termtrends canbe forecast,and behaviorand achievementscan be predicted
withfairaccuracyfrompreviousbehaviorsandachievements.Butweshouldnotexpect
performance in officer training and in combat to be predictable from behavior on an
obstacle field—behavior both on the test and in the real world is determined by many
factorsthatarespecifictotheparticularsituation.Removeonehighlyassertivemember
fromagroupofeightcandidatesandeveryoneelse’spersonalitieswillappeartochange.
Letasniper’sbulletmovebyafewcentimetersandtheperformanceofanofficerwillbe
transformed.Idonotdenythevalidityofalltests—ifatestpredictsanimportantoutcome
withavalidityof.20or.30,thetestshouldbeused.Butyoushouldnotexpectmore.You
should expect little or nothing from Wall Street stock pickers who hope to be more
accurate than the market in predicting the future of prices. And you should not expect
much from pundits making long-term forecasts—although they may have valuable
insightsintothenearfuture.Thelinethatseparatesthepossiblypredictablefuturefrom
theunpredictabledistantfutureisinyettobedrawn.
SpeakingofIllusorySkill
“He knows that the record indicates that the development of this illness is mostly
unpredictable.Howcanhebesoconfidentinthis case? Soundslikeanillusionof
validity.”
“Shehasacoherentstorythatexplainsallsheknows,andthecoherencemakesher
feelgood.”
“Whatmakeshimbelievethatheissmarterthanthemarket?Isthisanillusionof
skill?”
“Sheisahedgehog.Shehasatheorythatexplainseverything,anditgivesherthe
illusionthatsheunderstandstheworld.”
“Thequestionisnotwhethertheseexpertsarewelltrained.Itiswhethertheirworld
ispredictable.”
P
Intuitionsvs.Formulas
Paul Meehl was a strange and wonderful character, and one of the most versatile
psychologistsofthetwentieth century.Among the departmentsinwhichhe hadfaculty
appointmentsattheUniversityofMinnesotawerepsychology,law,psychiatry,neurology,
and philosophy. He also wrote on religion, political science, and learning in rats. A
statistically sophisticated researcher and a fierce critic of empty claims in clinical
psychology,Meehlwasalsoapracticingpsychoanalyst.Hewrotethoughtfulessaysonthe
philosophicalfoundationsofpsychologicalresearchthatIalmostmemorizedwhileIwas
agraduatestudent.InevermetMeehl,buthewasoneofmyheroesfromthetimeIread
his Clinical vs. Statistical Prediction: A Theoretical Analysis and a Review of the
Evidence.
Intheslimvolumethathelatercalled“mydisturbinglittlebook,”Meehlreviewed
the results of 20 studies that had analyzed whether clinical predictions based on the
subjective impressions of trained professionals were more accurate than statistical
predictionsmadebycombiningafewscoresorratingsaccordingtoarule.Inatypical
study,trainedcounselorspredictedthegradesoffreshmenattheendoftheschoolyear.
Thecounselorsinterviewedeachstudentforforty-fiveminutes.Theyalsohadaccessto
high school grades, several aptitude tests, and a four-page personal statement. The
statisticalalgorithmusedonlyafractionofthisinformation:highschoolgradesandone
aptitudetest.Nevertheless,theformulawasmoreaccuratethan11ofthe14counselors.
Meehl reported generally similar results across a variety of other forecast outcomes,
includingviolationsofparole,successinpilottraining,andcriminalrecidivism.
Not surprisingly, Meehl’s book provoked shock and disbelief among clinical
psychologists,andthecontroversyitstartedhasengenderedastreamofresearchthatis
stillflowingtoday,more than fiftyyephyЉdiars afteritspublication.Thenumber of
studies reporting comparisons of clinical and statistical predictions has increased to
roughlytwohundred,butthescoreinthecontestbetweenalgorithmsandhumanshasnot
changed. About 60% of the studies have shown significantly better accuracy for the
algorithms.Theothercomparisonsscoredadrawinaccuracy,butatieistantamounttoa
winforthestatisticalrules,whicharenormallymuchlessexpensive to use thanexpert
judgment.Noexceptionhasbeenconvincinglydocumented.
Therangeofpredictedoutcomeshasexpandedtocovermedicalvariablessuchasthe
longevityofcancerpatients,thelengthofhospitalstays,thediagnosisofcardiacdisease,
andthesusceptibilityofbabiestosuddeninfantdeathsyndrome;economicmeasuressuch
astheprospectsofsuccessfornewbusinesses,theevaluationofcreditrisksbybanks,and
the future career satisfaction of workers; questions of interest to government agencies,
including assessments of the suitability of foster parents, the odds of recidivism among
juvenile offenders, and the likelihood of other forms of violent behavior; and
miscellaneousoutcomessuchastheevaluationofscientificpresentations,thewinnersof
footballgames,andthefuturepricesofBordeauxwine.Eachofthesedomainsentailsa
significantdegreeofuncertaintyandunpredictability.Wedescribethemas“low-validity
environments.” In every case, the accuracy of experts was matched or exceeded by a
simplealgorithm.
As Meehl pointed out with justified pride thirty years after the publication of his
book, “There is no controversy in social science which shows such a large body of
qualitativelydiversestudiescomingoutsouniformlyinthesamedirectionasthisone.”
ThePrincetoneconomistandwineloverOrleyAshenfelterhasofferedacompelling
demonstration of the power of simple statistics to outdo world-renowned experts.
AshenfelterwantedtopredictthefuturevalueoffineBordeauxwinesfrominformation
available in the year they are made. The question is important because fine wines take
yearstoreachtheirpeakquality,andthepricesofmaturewinesfromthesamevineyard
vary dramatically across different vintages; bottles filled only twelve months apart can
differ in value by a factor of 10 or more. An ability to forecast future prices is of
substantialvalue,becauseinvestorsbuywine,likeart,intheanticipationthatitsvaluewill
appreciate.
Itisgenerallyagreedthat the effectof vintage can be due only to variationsinthe
weatherduringthegrape-growingseason.Thebestwinesareproducedwhenthesummer
iswarmanddry,whichmakestheBordeauxwineindustryalikelybeneficiaryofglobal
warming. The industry is also helped by wet springs, which increase quantity without
much effect on quality. Ashenfelter converted that conventional knowledge into a
statistical formula that predicts the price of a wine—for a particular property and at a
particularage—bythreefeaturesoftheweather:theaveragetemperatureoverthesummer
growing season, the amount of rain at harvest-time, and the total rainfall during the
previouswinter.Hisformulaprovidesaccuratepriceforecastsyearsandevendecadesinto
the future. Indeed, his formula forecasts future prices much more accurately than the
currentpricesofyoungwinesdo.Thisnewexampleofa“Meehlpattern”challengesthe
abilities of the experts whose opinions help shape the early price. It also challenges
economictheory,according towhich pricesshould reflect all the available information,
including the weather. Ashenfelter’s formula is extremely accurate—the correlation
betweenhispredictionsandactualpricesisabove.90.
Why are experts e yinferior to algorithms? One reason, which Meehl suspected, is
thatexpertstrytobeclever,thinkoutsidethebox,andconsidercomplexcombinationsof
featuresinmakingtheirpredictions.Complexitymayworkintheoddcase,butmoreoften
than not it reduces validity. Simple combinations of features are better. Several studies
haveshownthathumandecisionmakersareinferiortoapredictionformulaevenwhen
theyare giventhescoresuggestedbytheformula!Theyfeelthattheycanoverrulethe
formulabecausetheyhaveadditionalinformationaboutthecase,buttheyarewrongmore
oftenthannot.AccordingtoMeehl,therearefewcircumstancesunderwhichitisagood
ideatosubstitutejudgmentforaformula.Inafamousthoughtexperiment,hedescribeda
formulathatpredictswhetheraparticularpersonwillgotothemoviestonightandnoted
that it is proper to disregard the formula if information is received that the individual
broke a leg today. The name “broken-leg rule” has stuck. The point, of course, is that
brokenlegsareveryrare—aswellasdecisive.
Anotherreasonfortheinferiorityofexpertjudgmentisthathumansareincorrigibly
inconsistent in making summary judgments of complex information. When asked to
evaluatethesameinformationtwice,theyfrequentlygivedifferentanswers.Theextentof
theinconsistencyisoftenamatterofrealconcern.Experiencedradiologistswhoevaluate
chestX-raysas“normal”or“abnormal”contradictthemselves20%ofthetimewhenthey
seethesamepictureonseparateoccasions.Astudyof101independentauditorswhowere
askedtoevaluatethereliabilityofinternalcorporateauditsrevealedasimilardegreeof
inconsistency. A review of 41 separate studies of the reliability of judgments made by
auditors, pathologists, psychologists, organizational managers, and other professionals
suggeststhatthislevelofinconsistencyistypical,evenwhenacaseisreevaluatedwithin
afewminutes.Unreliablejudgmentscannotbevalidpredictorsofanything.
Thewidespreadinconsistencyisprobablyduetotheextremecontextdependencyof
System 1. We know from studies of priming that unnoticed stimuli in our environment
haveasubstantialinfluenceonourthoughtsandactions.Theseinfluencesfluctuatefrom
moment to moment. The brief pleasure of a cool breeze on a hot day may make you
slightlymorepositiveandoptimisticaboutwhateveryouareevaluatingatthetime.The
prospectsofaconvictbeinggrantedparolemaychangesignificantlyduringthetimethat
elapsesbetweensuccessivefoodbreaksintheparolejudges’schedule.Becauseyouhave
littledirectknowledgeofwhatgoesoninyourmind,youwillneverknowthatyoumight
have made a different judgment or reached a different decision under very slightly
different circumstances. Formulas do not suffer from such problems. Given the same
input, they always return the same answer. When predictability is poor—which it is in
mostofthestudiesreviewedbyMeehlandhisfollowers—inconsistencyisdestructiveof
anypredictivevalidity.
Theresearchsuggestsasurprisingconclusion:tomaximizepredictiveaccuracy,final
decisionsshouldbelefttoformulas,especiallyinlow-validityenvironments.Inadmission
decisionsformedicalschools,forexample,thefinaldeterminationisoftenmadebythe
facultymemberswhointerviewthecandidate.Theevidenceisfragmentary,butthereare
solidgroundsforaconjecture:conductinganinterviewislikelytodiminishtheaccuracy
of a selection procedure, if the interviewers also make the final admission decisions.
Because interviewers are overconfident in their intuitions, they will assign too much
weighttotheirpersonalimpressionsandtoolittleweighttoothersourcesofinformation,
loweringvalidity.Similarly,theexpertswhoevaluatethequasplityofimmaturewineto
predictitsfuturehaveasourceofinformationthatalmostcertainlymakesthingsworse
ratherthanbetter:theycantastethewine.Inaddition,ofcourse,eveniftheyhaveagood
understanding of the effects of the weather on wine quality, they will not be able to
maintaintheconsistencyofaformula.
The most important development in the field since Meehl’s original work is Robyn
Dawes’s famous article “The Robust Beauty of Improper Linear Models in Decision
Making.”Thedominantstatisticalpracticeinthesocialsciencesistoassignweightstothe
differentpredictorsbyfollowinganalgorithm,calledmultipleregression,thatisnowbuilt
intoconventional software. Thelogic ofmultiple regression isunassailable: itfinds the
optimalformulaforputtingtogetheraweightedcombinationofthepredictors.However,
Dawesobservedthatthecomplexstatisticalalgorithmaddslittleornovalue.Onecando
justaswellbyselectingasetofscoresthathavesomevalidityforpredictingtheoutcome
andadjustingthevaluestomakethemcomparable(byusingstandardscoresorranks).A
formulathatcombinesthesepredictorswithequalweightsislikelytobejustasaccurate
inpredictingnewcasesasthemultiple-regressionformulathatwasoptimalintheoriginal
sample.Morerecentresearchwentfurther:formulasthatassignequalweightstoallthe
predictorsareoftensuperior,becausetheyarenotaffectedbyaccidentsofsampling.
The surprising success of equal-weighting schemes has an important practical
implication: it is possible to develop useful algorithms without any prior statistical
research. Simple equally weighted formulas based on existing statistics or on common
sense are often very good predictors of significant outcomes. In a memorable example,
Dawesshowedthatmaritalstabilityiswellpredictedbyaformula:
frequencyoflovemakingminusfrequencyofquarrels
Youdon’twantyourresulttobeanegativenumber.
Theimportantconclusionfromthisresearchisthatanalgorithmthatisconstructedon
the back of an envelope is often good enough to compete with an optimally weighted
formula,andcertainlygoodenoughtooutdoexpertjudgment.Thislogiccanbeappliedin
manydomains,rangingfromtheselectionofstocksbyportfoliomanagerstothechoices
ofmedicaltreatmentsbydoctorsorpatients.
Aclassicapplicationofthisapproachisasimplealgorithmthathassavedthelivesof
hundredsofthousandsofinfants.Obstetricianshadalwaysknownthataninfantwhois
notbreathingnormallywithinafewminutesofbirthisathighriskofbraindamageor
death. Until the anesthesiologist Virginia Apgar intervened in 1953, physicians and
midwives used their clinical judgment to determine whether a baby was in distress.
Different practitioners focused on different cues. Some watched for breathing problems
while others monitored how soon the baby cried. Without a standardized procedure,
dangersignswereoftenmissed,andmanynewborninfantsdied.
Onedayoverbreakfast,amedicalresidentaskedhowDr.Apgarwouldmakeasystematic
assessmentofanewborn.“That’seasy,”shereplied.“Youwoulddoitlikethis.”Apgar
jotteddownfivevariables(heartrate,respiration,reflex,muscletone,andcolor)andthree
scores(0,1,or2,dependingontherobustnessofeachsign).Realizingthatshemighthave
made a breakequthrough that any delivery room could implement, Apgar began rating
infants by this rule one minute after they were born. A baby with a total score of 8 or
abovewaslikelytobepink,squirming,crying,grimacing,withapulseof100ormore—
ingoodshape.Ababywithascoreof4orbelowwasprobablybluish,flaccid,passive,
withasloworweakpulse—inneedofimmediateintervention.ApplyingApgar’sscore,
thestaffindeliveryroomsfinallyhadconsistentstandardsfordeterminingwhichbabies
were in trouble, and the formula is credited for an important contribution to reducing
infant mortality. The Apgar test is still used every day in every delivery room. Atul
Gawande’srecentAChecklistManifestoprovidesmanyotherexamplesofthevirtuesof
checklistsandsimplerules.
TheHostilitytoAlgorithms
Fromtheveryoutset,clinicalpsychologistsrespondedtoMeehl’sideaswithhostilityand
disbelief.Clearly,theywereinthegripofanillusionofskillintermsoftheirabilityto
makelong-termpredictions.Onreflection,itiseasytoseehowtheillusioncameabout
andeasytosympathizewiththeclinicians’rejectionofMeehl’sresearch.
The statistical evidence of clinical inferiority contradicts clinicians’ everyday
experienceofthequalityoftheirjudgments.Psychologistswhoworkwithpatientshave
manyhunchesduringeachtherapysession,anticipatinghowthepatientwillrespondtoan
intervention, guessing what will happen next. Many of these hunches are confirmed,
illustratingtherealityofclinicalskill.
The problem is that the correct judgments involve short-term predictions in the
contextofthetherapeuticinterview,askillinwhichtherapistsmayhaveyearsofpractice.
The tasks at which they fail typically require long-term predictions about the patient’s
future.Thesearemuchmoredifficult,eventhebestformulasdoonlymodestlywell,and
theyarealsotasksthattheclinicianshaveneverhadtheopportunitytolearnproperly—
they would have to wait years for feedback, instead of receiving the instantaneous
feedbackoftheclinicalsession.However,the line betweenwhatclinicianscandowell
andwhattheycannotdoatallwellisnotobvious,andcertainlynotobvioustothem.They
knowtheyareskilled,buttheydon’tnecessarilyknowtheboundariesoftheirskill.Not
surprisingly, then, the idea that a mechanical combination of a few variables could
outperform the subtle complexity of human judgment strikes experienced clinicians as
obviouslywrong.
The debate about the virtues of clinical and statistical prediction has always had a
moral dimension. The statistical method, Meehl wrote, was criticized by experienced
clinicians as “mechanical, atomistic, additive, cut and dried, artificial, unreal, arbitrary,
incomplete, dead, pedantic, fractionated, trivial, forced, static, superficial, rigid, sterile,
academic,pseudoscientificandblind.”Theclinicalmethod,ontheotherhand,waslauded
by its proponents as “dynamic, global, meaningful, holistic, subtle, sympathetic,
configural,patterned,organized,rich,deep,genuine,sensitive,sophisticated,real,living,
concrete,natural,truetolife,andunderstanding.”
Thisis anattitudewe canall recognize.When a humancompetes witha machine,
whether it is John Henry a-hammerin’ on the mountain or the chess genius Garry
KasparovfacingoffagainstthecomputerDeepBlue,oursympathiesliewithourfellow
human.Theaversiontoalgorithmsmakingdecisionsthataffecthumansisrootedinthe
strongpreferencethatmanypeoplehavefortheormnaturaloverthesyntheticorartificial.
Asked whether they would rather eat an organic or a commercially grown apple, most
peoplepreferthe“allnatural”one.Evenafterbeinginformedthatthetwoapplestastethe
same,haveidenticalnutritionalvalue,andareequallyhealthful,amajoritystillpreferthe
organicfruit.Eventheproducersofbeerhavefoundthattheycanincreasesalesbyputting
“AllNatural”or“NoPreservatives”onthelabel.
Thedeepresistancetothedemystificationofexpertiseisillustratedbythereactionof
the European wine community to Ashenfelter’s formula for predicting the price of
Bordeauxwines.Ashenfelter’sformulaansweredaprayer:onemightthushaveexpected
thatwineloverseverywherewouldbegratefultohimfordemonstrablyimprovingtheir
abilitytoidentifythewinesthatlaterwouldbegood.Notso.TheresponseinFrenchwine
circles,wroteTheNewYorkTimes,ranged“somewherebetweenviolentandhysterical.”
Ashenfelterreportsthatoneoenophilecalledhisfindings“ludicrousandabsurd.”Another
scoffed,“Itislikejudgingmovieswithoutactuallyseeingthem.”
Theprejudiceagainstalgorithmsismagnifiedwhenthedecisionsareconsequential.
Meehlremarked,“Idonotquiteknowhowtoalleviatethehorrorsomecliniciansseemto
experiencewhentheyenvisageatreatablecasebeingdeniedtreatmentbecausea‘blind,
mechanical’ equation misclassifies him.” In contrast, Meehl and other proponents of
algorithms have argued strongly that it is unethical to rely on intuitive judgments for
important decisions if an algorithm is available that will make fewer mistakes. Their
rationalargumentiscompelling,butitrunsagainstastubbornpsychologicalreality:for
most people, the cause of a mistake matters. The story of a child dying because an
algorithmmadeamistakeismorepoignantthanthestoryofthesametragedyoccurringas
aresultofhumanerror,andthedifferenceinemotionalintensityisreadilytranslatedinto
amoralpreference.
Fortunately,thehostilitytoalgorithmswillprobablysoftenastheirroleineveryday
life continues to expand. Looking for books or music we might enjoy, we appreciate
recommendationsgeneratedbysoftware.Wetakeitforgrantedthatdecisionsaboutcredit
limits are made without the direct intervention of any human judgment. We are
increasinglyexposedtoguidelinesthathave the formofsimplealgorithms,suchas the
ratioofgoodandbadcholesterollevelsweshouldstrivetoattain.Thepublicisnowwell
awarethatformulasmaydobetterthanhumansinsomecriticaldecisionsintheworldof
sports:howmuchaprofessionalteamshouldpayforparticularrookieplayers,orwhento
puntonfourthdown.Theexpandinglistoftasksthatareassignedtoalgorithmsshould
eventually reduce the discomfort that most people feel when they first encounter the
patternofresultsthatMeehldescribedinhisdisturbinglittlebook.
LearningfromMeehl
In1955,asatwenty-one-year-oldlieutenantintheIsraeliDefenseForces,Iwasassigned
tosetupaninterviewsystemfortheentirearmy.Ifyouwonderwhysucharesponsibility
wouldbeforceduponsomeonesoyoung,bearinmindthatthestateofIsraelitselfwas
onlysevenyearsoldatthetime;allitsinstitutionswereunderconstruction,andsomeone
hadtobuildthem.Oddasitsoundstoday,mybachelor’sdegreeinpsychologyprobably
qualifiedmeasthebest-trainedpsychologistinthearmy.Mydirectsupervisor,abrilliant
researcher,hadadegreeinchemistry.
An idilnterview routine was already in place when I was given my mission. Every
soldier drafted into the army completed a battery of psychometric tests, and each man
considered for combat duty was interviewed for an assessment of personality. The goal
wastoassigntherecruitascoreofgeneralfitnessforcombatandtofindthebestmatchof
his personality among various branches: infantry, artillery, armor, and so on. The
interviewers were themselves young draftees, selected for this assignment by virtue of
theirhighintelligenceandinterestindealingwithpeople.Mostwerewomen,whowereat
thetimeexemptfromcombatduty.Trainedforafewweeksinhowtoconductafifteen-to
twenty-minuteinterview,theywereencouragedtocoverarangeoftopicsandtoforma
generalimpressionofhowwelltherecruitwoulddointhearmy.
Unfortunately, follow-up evaluations had already indicated that this interview
procedurewasalmostuselessforpredictingthefuturesuccessofrecruits.Iwasinstructed
todesignaninterviewthatwouldbemoreusefulbutwouldnottakemoretime.Iwasalso
toldtotryoutthenewinterviewandtoevaluateitsaccuracy.Fromtheperspectiveofa
serious professional, I was no more qualified for the task than I was to build a bridge
acrosstheAmazon.
Fortunately, I had read Paul Meehl’s “little book,” which had appeared just a year
earlier. I was convinced by his argument that simple, statistical rules are superior to
intuitive “clinical” judgments. I concluded that the then current interview had failed at
leastinpartbecauseitallowedtheinterviewerstodowhattheyfoundmostinteresting,
whichwastolearnaboutthedynamicsoftheinterviewee’smentallife.Instead,weshould
use the limited time at our disposal to obtain as much specific information as possible
about the interviewee’s life in his normal environment. Another lesson I learned from
Meehl was that we should abandon the procedure in which the interviewers’ global
evaluationsoftherecruitdeterminedthefinaldecision.Meehl’sbooksuggestedthatsuch
evaluations should not be trusted and that statistical summaries of separately evaluated
attributeswouldachievehighervalidity.
Idecidedonaprocedureinwhichtheinterviewerswouldevaluateseveralrelevant
personality traits and score each separately. The final score of fitness for combat duty
would be computed according to a standard formula, with no further input from the
interviewers.Imadeupalistofsixcharacteristicsthatappearedrelevanttoperformance
inacombat unit,including“responsibility,”“sociability,” and“masculinepride.” I then
composed,foreachtrait,aseriesoffactualquestionsabouttheindividual’slifebeforehis
enlistment,includingthenumberofdifferentjobshehadheld,howregularandpunctual
hehadbeeninhisworkorstudies,thefrequencyofhisinteractionswithfriends,andhis
interestandparticipationinsports,amongothers.Theideawastoevaluateasobjectively
aspossiblehowwelltherecruithaddoneoneachdimension.
By focusing on standardized, factual questions, I hoped to combat the halo effect,
where favorable first impressions influence later judgments. As a further precaution
againsthalos,Iinstructedtheinterviewerstogothroughthesixtraitsinafixedsequence,
rating each trait on a five-point scale before going on to the next. And that was that. I
informedtheinterviewersthattheyneednotconcernthemselveswiththerecruit’sfuture
adjustmenttothemilitary.Theironlytaskwastoelicitrelevantfactsabouthispastandto
use that information to score each personality dimension. “Your function is to provide
reliablemeasurements,”Itoldthem.“Leavethepredicoktivevaliditytome,”bywhichI
meanttheformulathatIwasgoingtodevisetocombinetheirspecificratings.
Theinterviewerscameclosetomutiny.Thesebrightyoungpeopleweredispleasedto
be ordered, by someone hardly older than themselves, to switch off their intuition and
focusentirelyonboringfactualquestions.Oneofthemcomplained,“Youareturningus
into robots!” So I compromised. “Carry out the interview exactly as instructed,” I told
them,“andwhenyouaredone,haveyourwish:closeyoureyes,trytoimaginetherecruit
asasoldier,andassignhimascoreonascaleof1to5.”
Severalhundredinterviewswereconductedbythisnewmethod,andafewmonths
laterwecollectedevaluationsofthesoldiers’performancefromthecommandingofficers
oftheunitstowhichtheyhadbeenassigned.Theresultsmadeushappy.AsMeehl’sbook
hadsuggested,thenewinterviewprocedurewasasubstantialimprovementovertheold
one. The sum of our six ratings predicted soldiers’ performance much more accurately
than the global evaluations of the previous interviewing method, although far from
perfectly.Wehadprogressedfrom“completelyuseless”to“moderatelyuseful.”
The big surprise to me was that the intuitive judgment that the interviewers
summonedupinthe“closeyoureyes”exercisealsodidverywell,indeedjustaswellas
thesumofthesixspecificratings.IlearnedfromthisfindingalessonthatIhavenever
forgotten:intuitionaddsvalueeveninthejustlyderidedselectioninterview,butonlyafter
adisciplinedcollectionofobjectiveinformationanddisciplinedscoringofseparatetraits.
Isetaformulathatgavethe“closeyoureyes”evaluationthesameweightasthesumof
the six trait ratings. A more general lesson that I learned from this episode was do not
simplytrustintuitivejudgment—yourownorthatofothers—butdonotdismissit,either.
Someforty-fiveyearslater,afterIwonaNobelPrizeineconomics,Iwasforashort
timeaminorcelebrityinIsrael.Ononeofmyvisits,someonehadtheideaofescortingme
aroundmyoldarmybase,whichstillhousedtheunitthatinterviewsnewrecruits.Iwas
introducedtothecommandingofficerofthePsychologicalUnit,andshedescribedtheir
current interviewing practices, which had not changed much from the system I had
designed;therewas,itturnedout,aconsiderableamountofresearchindicatingthatthe
interviews still worked well. As she came to the end of her description of how the
interviewsareconducted,theofficeradded,“Andthenwetellthem,‘Closeyoureyes.’”
DoItYourself
Themessageof thischapteris readilyapplicableto tasksotherthan makingmanpower
decisions for an army. Implementing interview procedures in the spirit of Meehl and
Dawesrequiresrelativelylittleeffortbutsubstantialdiscipline.Supposethatyouneedto
hireasalesrepresentativeforyourfirm.Ifyouareseriousabouthiringthebestpossible
person for the job, this is what you should do. First, select a few traits that are
prerequisites for success in this position (technical proficiency, engaging personality,
reliability,andsoon).Don’toverdoit—sixdimensionsisagoodnumber.Thetraitsyou
chooseshouldbeasindependentaspossiblefromeachother,andyoushouldfeelthatyou
can assess them reliably by asking a few factual questions. Next, make a list of those
questionsfor eachtrait andthink abouthow youwill scoreit, sayon a1–5 scale.You
shouldhaveanideaofwhatyouwillcaleigl“veryweak”or“verystrong.”
Thesepreparationsshould take youhalfan hour orso,a smallinvestmentthatcan
makeasignificantdifferenceinthequalityofthepeopleyouhire.Toavoidhaloeffects,
youmustcollecttheinformationononetraitatatime,scoringeachbeforeyoumoveon
tothenextone.Donotskiparound.Toevaluateeachcandidate,addupthesixscores.
Becauseyouareinchargeofthefinaldecision,youshouldnotdoa“closeyoureyes.”
Firmlyresolvethatyouwill hire thecandidatewhosefinalscore isthehighest,evenif
thereisanotheronewhomyoulikebetter—trytoresistyourwishtoinventbrokenlegsto
changetheranking.Avastamountofresearchoffersapromise:youaremuchmorelikely
tofindthebestcandidateifyouusethisprocedurethanifyoudowhatpeoplenormallydo
insuchsituations,whichistogointotheinterviewunpreparedandtomakechoicesbyan
overallintuitivejudgmentsuchas“IlookedintohiseyesandlikedwhatIsaw.”
SpeakingofJudgesvs.Formulas
“Whenever we can replace human judgment by a formula, we should at least
considerit.”
“Hethinkshisjudgmentsarecomplexandsubtle,butasimplecombinationofscores
couldprobablydobetter.”
“Let’sdecideinadvancewhatweighttogivetothedatawehaveonthecandidates’
pastperformance.Otherwisewewillgivetoomuchweighttoourimpressionfrom
theinterviews.”
P
ExpertIntuition:WhenCanWeTrustIt?
Professional controversies bring out the worst in academics. Scientific journals
occasionally publish exchanges, often beginning with someone’s critique of another’s
research,followedbyareplyandarejoinder.Ihavealwaysthoughtthattheseexchanges
areawasteoftime.Especiallywhentheoriginalcritiqueissharplyworded,thereplyand
therejoinderareoftenexercisesinwhatIhavecalledsarcasmforbeginnersandadvanced
sarcasm.Therepliesrarelyconcedeanythingtoabitingcritique,anditisalmostunheard
offor arejoinder toadmit thatthe originalcritique wasmisguided orerroneous inany
way. On a few occasions I have responded to criticisms that I thought were grossly
misleading,becauseafailuretorespondcanbeinterpretedasconcedingerror,butIhave
never found the hostile exchanges instructive. In search of another way to deal with
disagreements, I have engaged in a few “adversarial collaborations,” in which scholars
whodisagreeonthescienceagreetowriteajointlyauthoredpaperontheirdifferences,
and sometimes conduct research together. In especially tense situations, the research is
moderatedbyanarbiter.
MymostsatisfyingandproductiveadversarialcollaborationwaswithGaryKlein,the
intellectualleaderofanassociationofscholarsandpractitionerswhodonotlikethekind
ofworkIdo. They callthemselvesstudentsofNaturalisticDecisionMaking,or NDM,
andmostlyworkinorganizationswherethe”0%Љtyoftenstudyhowexpertswork.The
NDMersadamantlyrejectthefocusonbiasesintheheuristicsandbiasesapproach.They
criticizethismodelasoverlyconcernedwithfailuresanddrivenbyartificialexperiments
ratherthanbythestudyofrealpeopledoingthingsthatmatter.Theyaredeeplyskeptical
aboutthevalueofusingrigidalgorithmstoreplacehumanjudgment,andPaulMeehlis
not among their heroes. Gary Klein has eloquently articulated this position over many
years.
Thisishardlythebasisforabeautifulfriendship,butthereismoretothestory.Ihad
neverbelievedthatintuitionisalwaysmisguided.IhadalsobeenafanofKlein’sstudies
ofexpertiseinfirefighterssinceIfirstsawadraftofapaperhewroteinthe1970s,and
wasimpressedbyhisbookSourcesofPower,muchofwhichanalyzeshowexperienced
professionalsdevelopintuitiveskills.Iinvitedhimtojoininanefforttomaptheboundary
thatseparatesthemarvelsofintuitionfromitsflaws.Hewasintriguedbytheideaandwe
wentaheadwiththeproject—withnocertaintythatitwouldsucceed.Wesetouttoanswer
aspecificquestion:Whencanyoutrustanexperiencedprofessionalwhoclaimstohave
anintuition?ItwasobviousthatKleinwouldbemoredisposedtobetrusting,andIwould
bemoreskeptical.Butcouldweagreeonprinciplesforansweringthegeneralquestion?
Oversevenor eightyearswe hadmanydiscussions, resolvedmanydisagreements,
almost blew up more than once, wrote many draft s, became friends, and eventually
publishedajointarticlewithatitlethattellsthestory:“ConditionsforIntuitiveExpertise:
AFailuretoDisagree.”Indeed,wedidnotencounterrealissuesonwhichwedisagreed—
butwedidnotreallyagree.
MarvelsandFlaws
Malcolm Gladwell’s bestseller Blink appeared while Klein and I were working on the
project, and it was reassuring to find ourselves in agreement about it. Gladwell’s book
openswiththememorablestoryofartexpertsfacedwithanobjectthatisdescribedasa
magnificentexampleofakouros,asculptureofastridingboy.Severaloftheexpertshad
strongvisceralreactions:theyfeltintheirgutthatthestatuewasafakebutwerenotable
toarticulatewhatitwasaboutitthatmadethemuneasy.Everyonewhoreadthebook—
millionsdid—remembersthatstoryasatriumphofintuition.Theexpertsagreedthatthey
knewthesculpturewasafakewithoutknowinghowtheyknew—theverydefinitionof
intuition.Thestoryappearstoimplythatasystematicsearchforthecuethatguidedthe
expertswouldhavefailed,butKleinandIbothrejectedthatconclusion.Fromourpointof
view, such an inquiry was needed, and if it had been conducted properly (which Klein
knowshowtodo),itwouldprobablyhavesucceeded.
Although many readers of the kouros example were surely drawn to an almost
magicalviewofexpertintuition,Gladwellhimselfdoesnotholdthatposition.Inalater
chapterhedescribesamassivefailureofintuition:AmericanselectedPresidentHarding,
whose only qualification for the position was that he perfectly looked the part. Square
jawedandtall,hewastheperfectimageofastronganddecisiveleader.Peoplevotedfor
someonewholookedstronganddecisivewithoutanyotherreasontobelievethathewas.
An intuitive prediction of how Harding would perform as president arose from
substitutingonequestionforanother.Areaderofthisbookshouldexpectsuchanintuition
tobeheldwithconfidence.
IntuitionasRecognition
TheearlyexperiencesthatshapedKlein’sviewsofintuitionwerestarklydifferentfrom
mine. My thinking was formed by observing the illusion of validity in myself and by
readingPaulMeehl’sdemonstrationsoftheinferiorityofclinicalprediction.Incontrast,
Klein’sviewswereshapedbyhisearlystudiesoffiregroundcommanders(theleadersof
firefightingteams).Hefollowedthemastheyfoughtfiresandlaterinterviewedtheleader
abouthisthoughtsashemadedecisions.AsKleindescribeditinourjointarticle,heand
hiscollaborators
investigated how the commanders could make good decisions without comparing
options.Theinitialhypothesiswasthatcommanderswouldrestricttheiranalysisto
only a pair of options, but that hypothesis proved to be incorrect. In fact, the
commanders usually generated only a single option, and that was all they needed.
Theycould drawon therepertoire ofpatterns thatthey had compiled duringmore
thanadecadeofbothrealandvirtualexperiencetoidentifyaplausibleoption,which
theyconsideredfirst.Theyevaluatedthisoptionbymentallysimulatingittoseeifit
would work in the situation they were facing…. If the course of action they were
considering seemed appropriate, they would implement it. If it had shortcomings,
theywouldmodifyit.Iftheycouldnoteasilymodifyit,theywouldturntothenext
mostplausibleoptionandrunthroughthesameprocedureuntilanacceptablecourse
ofactionwasfound.
Klein elaborated this description into a theory of decision making that he called the
recognition-primeddecision(RPD)model,whichappliestofirefightersbutalsodescribes
expertise in other domains, including chess. The process involves both System 1 and
System2.Inthefirstphase,atentativeplancomestomindbyanautomaticfunctionof
associativememory—System1.Thenextphaseisadeliberateprocessinwhichtheplanis
mentally simulated to check if it will work—an operation of System 2. The model of
intuitivedecisionmakingaspatternrecognitiondevelopsideaspresentedsometimeago
byHerbertSimon,perhapstheonlyscholarwhoisrecognizedandadmiredasaheroand
foundingfigurebyallthecompetingclansandtribesinthestudyofdecisionmaking.I
quotedHerbertSimon’sdefinitionofintuitionintheintroduction,butitwillmakemore
sense when I repeat it now: “The situation has provided a cue; this cue has given the
expertaccesstoinformationstoredinmemory,andtheinformationprovidestheanswer.
Intuitionisnothingmoreandnothinglessthanrecognition.”
This strong statement reduces the apparent magic of intuition to the everyday
experienceofmemory.Wemarvelatthestoryofthefirefighterwhohasasuddenurgeto
escapeaburninghousejustbeforeitcollapses,becausethefirefighterknowsthedanger
intuitively, “without knowing how he knows.” However, we also do not know how we
immediatelyknowthatapersonweseeasweenteraroomisourfriendPeter.Themoral
of Simon’s remark is that the mystery of knowing without knowing is not a distinctive
featureofintuition;itisthenormofmentallife.
AcquiringSkill
Howdoestheinformationthatsupportsintuitionget“storedinmemory”?Certaintypesof
intuitionsareacquiredveryquickly.Wehaveinheritedfromourancestorsagreatfacility
tolearnwhentobeafraid.Indeed,oneexperienceisoftensufficienttoestablishalong-
termaversionandfear.Manyofushavethevisceralmemoryofasingledubiousdishtto
hatstillleavesusvaguelyreluctanttoreturntoarestaurant.Allofustenseupwhenwe
approachaspotinwhichanunpleasanteventoccurred,evenwhenthereisnoreasonto
expectittohappenagain.Forme,onesuchplaceistherampleadingtotheSanFrancisco
airport,whereyearsagoadriverinthethroesofroadragefollowedmefromthefreeway,
rolled down his window, and hurled obscenities at me. I never knew what caused his
hatred,butIrememberhisvoicewheneverIreachthatpointonmywaytotheairport.
Mymemoryoftheairportincidentisconsciousanditfullyexplainstheemotionthat
comeswithit.Onmanyoccasions,however,youmayfeeluneasyinaparticularplaceor
whensomeoneusesaparticularturnofphrasewithouthavingaconsciousmemoryofthe
triggeringevent.Inhindsight,youwilllabelthatuneaseanintuitionifitisfollowedbya
badexperience.Thismodeofemotionallearningiscloselyrelatedtowhathappenedin
Pavlov’s famous conditioning experiments, in which the dogs learned to recognize the
soundofthebellasasignalthatfoodwascoming.WhatPavlov’s dogslearnedcanbe
describedasalearnedhope.Learnedfearsareevenmoreeasilyacquired.
Fearcanalsobelearned—quiteeasily,infact—bywordsratherthanbyexperience.
The fireman who had the “sixth sense” of danger had certainly had many occasions to
discussandthinkabouttypesoffireshewasnotinvolvedin,andtorehearseinhismind
whatthecuesmightbeandhowheshouldreact.AsIrememberfromexperience,ayoung
platoon commander with no experience of combat will tense up while leading troops
throughanarrowingravine,becausehewastaughttoidentifytheterrainasfavoringan
ambush.Littlerepetitionisneededforlearning.
Emotionallearningmaybequick,butwhatweconsideras“expertise”usuallytakesa
long time to develop. The acquisition of expertise in complex tasks such as high-level
chess,professionalbasketball,orfirefightingisintricateandslowbecauseexpertiseina
domain is not a single skill but rather a large collection of miniskills. Chess is a good
example. An expert player can understand a complex position at a glance, but it takes
years to develop that level of ability. Studies of chess masters have shown that at least
10,000 hours of dedicated practice (about 6 years of playing chess 5 hours a day) are
required to attain the highest levels of performance. During those hours of intense
concentration,aseriouschessplayerbecomesfamiliarwiththousandsofconfigurations,
eachconsistingofanarrangementofrelatedpiecesthatcanthreatenordefendeachother.
Learninghigh-levelchesscanbecomparedtolearningtoread.Afirstgraderworks
hardatrecognizingindividuallettersandassemblingthemintosyllablesandwords,buta
goodadultreaderperceivesentireclauses.Anexpertreaderhasalsoacquiredtheabilityto
assemble familiar elements in a new pattern and can quickly “recognize” and correctly
pronounceawordthatshehasneverseenbefore.Inchess,recurrentpatternsofinteracting
piecesplaytheroleofletters,andachesspositionisalongwordorasentence.
Askilledreaderwhoseesitforthefirsttimewillbeabletoreadtheopeningstanzaof
LewisCarroll’s“Jabberwocky”withperfectrhythmandintonation,aswellaspleasure:
’Twasbrillig,andtheslithytoves
Didgyreandgimbleinthewabe:
Allmimsyweretheborogoves,
Andthemomerathsoutgrabe.
Acquiringexpertiseinchessisharderandslowerthanlearningtoreadbecausethereare
many more letters in the “alphabet” of chess and because the “words” consist of many
letters. After thousands of hours of practice, however, chess masters are able to read a
chess situation at a glance. The few moves that come to their mind are almost always
strongandsometimescreative.Theycandealwitha“word”theyhaveneverencountered,
andtheycanfindanewwaytointerpretafamiliarone.
TheEnvironmentofSkill
KleinandIquicklyfoundthatweagreedbothonthenatureofintuitiveskillandonhowit
is acquired. We still needed to agree on our key question: When can you trust a self-
confidentprofessionalwhoclaimstohaveanintuition?
Weeventuallyconcludedthatourdisagreementwasdueinparttothefactthatwehad
different experts in mind. Klein had spent much time with fireground commanders,
clinical nurses, and other professionals who have real expertise. I had spent more time
thinking about clinicians, stock pickers, and political scientists trying to make
unsupportable long-term forecasts. Not surprisingly, his default attitude was trust and
respect;minewasskepticism.Hewasmorewillingtotrustexpertswhoclaimanintuition
because,ashetoldme,trueexpertsknowthelimitsoftheirknowledge.Iarguedthatthere
aremanypseudo-expertswhohavenoideathattheydonotknowwhattheyaredoing(the
illusionofvalidity),andthatasageneralpropositionsubjectiveconfidenceiscommonly
toohighandoftenuninformative.
EarlierItracedpeople’sconfidenceinabelieftotworelatedimpressions:cognitive
easeand coherence. We are confident when thestory we tell ourselves comeseasily to
mind,withnocontradictionandno competingscenario.Buteaseandcoherencedo not
guarantee that a belief held with confidence is true. The associative machine is set to
suppressdoubtandtoevokeideasandinformationthatarecompatiblewiththecurrently
dominantstory. A mindthatfollowsWYSIATIwillachievehighconfidencemuchtoo
easilybyignoringwhatitdoesnotknow.Itisthereforenotsurprisingthatmanyofusare
pronetohavehighconfidenceinunfoundedintuitions.KleinandIeventuallyagreedon
animportantprinciple:theconfidencethatpeoplehaveintheirintuitionsisnotareliable
guide to their validity. In other words, do not trust anyone—including yourself—to tell
youhowmuchyoushouldtrusttheirjudgment.
If subjective confidence is not to be trusted, how can we evaluate the probable
validity of an intuitive judgment? When do judgments reflect true expertise? When do
theydisplayanillusionofvalidity?Theanswercomesfromthetwobasicconditionsfor
acquiringaskill:
anenvironmentthatissufficientlyregulartobepredictable
anopportunitytolearntheseregularitiesthroughprolongedpractice
Whenboth theseconditionsare satisfied,intuitions are likelyto beskilled. Chessis an
extreme example of a regular environment, but bridge and poker also provide robust
statisticalregularitiesthatcansupportskill. Physicians, nurses,athletes,andfirefighters
alsofacecomplexbutfundamentallyorderlysituations.TheaccurateintuitionsthatGary
Kleinhasdescribedareduetohighlyvalidcuesthatestheexpert’sSystem1haslearned
to use, even if System 2 has not learned to name them. In contrast, stock pickers and
politicalscientistswhomakelong-termforecastsoperateinazero-validityenvironment.
Theirfailuresreflectthebasicunpredictabilityoftheeventsthattheytrytoforecast.
Some environments are worse than irregular. Robin Hogarth described “wicked”
environments, in which professionals are likely to learn the wrong lessons from
experience. He borrows from Lewis Thomas the example of a physician in the early
twentieth century who often had intuitions about patients who were about to develop
typhoid. Unfortunately, he tested his hunch by palpating the patient’s tongue, without
washinghishandsbetweenpatients.Whenpatientafterpatientbecameill,thephysician
developedasenseofclinicalinfallibility.Hispredictionswereaccurate—butnotbecause
hewasexercisingprofessionalintuition!
Meehl’s clinicians were not inept and their failure was not due to lack of talent. They
performedpoorlybecause they were assigned tasksthatdidnothaveasimplesolution.
Theclinicians’predicamentwaslessextremethanthezero-validityenvironmentoflong-
termpoliticalforecasting,buttheyoperatedinlow-validitysituationsthatdidnotallow
high accuracy. We know this to be the case because the best statistical algorithms,
althoughmoreaccuratethanhumanjudges,wereneververyaccurate.Indeed,thestudies
by Meehl and his followers never produced a “smoking gun” demonstration, a case in
which clinicians completely missed a highly valid cue that the algorithm detected. An
extremefailureofthiskindisunlikelybecausehumanlearningisnormallyefficient.Ifa
strongpredictivecueexists,humanobserverswillfindit,givenadecentopportunitytodo
so. Statistical algorithms greatly outdo humans in noisy environments for two reasons:
theyaremorelikelythanhumanjudgestodetectweaklyvalidcuesandmuchmorelikely
tomaintainamodestlevelofaccuracybyusingsuchcuesconsistently.
It is wrong to blame anyone for failing to forecast accurately in an unpredictable
world.However,itseemsfairtoblameprofessionalsforbelievingtheycansucceedinan
impossible task. Claims for correct intuitions in an unpredictable situation are self-
delusionalatbest,sometimesworse.Intheabsenceofvalidcues,intuitive“hits”aredue
eithertoluckortolies.Ifyoufindthisconclusionsurprising,youstillhavealingering
belief that intuition is magic. Remember this rule: intuition cannot be trusted in the
absenceofstableregularitiesintheenvironment.
FeedbackandPractice
Someregularitiesintheenvironmentareeasiertodiscoverandapplythanothers.Thinkof
howyoudevelopedyourstyleofusingthebrakesonyourcar.Asyouweremasteringthe
skilloftakingcurves,yougraduallylearnedwhentoletgooftheacceleratorandwhen
andhowhardtousethebrakes.Curvesdiffer,andthevariabilityyouexperiencedwhile
learningensures that you are nowready tobrake atthe right timeand strengthfor any
curveyouencounter.Theconditionsforlearningthisskillareideal,becauseyoureceive
immediateandunambiguousfeedbackeverytimeyougoaroundabend:themildreward
ofacomfortableturnorthemildpunishmentofsomedifficultyinhandlingthecarifyou
brake either too hard or not quite hard enough. The situations that face a harbor pilot
maneuveringlargeshipsarenolessregular,butskillismuchmoredifficulttoacquireby
sheer experience because of the long delay between actions and their manoticeable
outcomes. Whether professionals have a chance to develop intuitive expertise depends
essentiallyonthequalityandspeedoffeedback,aswellasonsufficientopportunityto
practice.
Expertiseisnota single skill;itisacollectionof skills,andthesameprofessional
may be highly expert in some of the tasks in her domain while remaining a novice in
others.Bythetimechessplayersbecomeexperts,theyhave“seeneverything”(oralmost
everything), but chess is an exception in this regard. Surgeons can be much more
proficient in some operations than in others. Furthermore, some aspects of any
professional’s tasks are much easier to learn than others. Psychotherapists have many
opportunities to observe the immediate reactions of patients to what they say. The
feedbackenablesthemtodeveloptheintuitiveskilltofindthewordsandthetonethatwill
calmanger,forgeconfidence,orfocusthepatient’sattention.Ontheotherhand,therapists
do not have a chance to identify which general treatment approach is most suitable for
different patients. The feedback they receive from their patients’ long-term outcomes is
sparse, delayed, or (usually) nonexistent, and in any case too ambiguous to support
learningfromexperience.
Among medical specialties, anesthesiologists benefit from good feedback, because
theeffectsoftheiractionsarelikelytobequicklyevident.Incontrast,radiologistsobtain
littleinformationabouttheaccuracyofthediagnosestheymakeandaboutthepathologies
theyfail todetect. Anesthesiologistsare thereforein abetter positionto developuseful
intuitive skills. If an anesthesiologist says, “I have a feeling something is wrong,”
everyoneintheoperatingroomshouldbepreparedforanemergency.
Here again, as in the case of subjective confidence, the experts may not know the
limits of their expertise. An experienced psychotherapist knows that she is skilled in
workingoutwhatisgoingoninherpatient’smindandthatshehasgoodintuitionsabout
what the patient will say next. It is tempting for her to conclude that she can also
anticipate how well the patient will do next year, but this conclusion is not equally
justified. Short-term anticipation and long-term forecasting are different tasks, and the
therapisthashadadequateopportunitytolearnonebutnottheother.Similarly,afinancial
expertmayhaveskillsinmanyaspectsofhistradebutnotinpickingstocks,andanexpert
intheMiddleEastknowsmanythingsbutnotthefuture.Theclinicalpsychologist,the
stockpicker,andthepunditdohaveintuitiveskillsinsomeoftheirtasks,buttheyhave
notlearnedtoidentifythesituationsandthetasksinwhichintuitionwillbetraythem.The
unrecognizedlimitsofprofessionalskillhelpexplainwhyexpertsareoftenoverconfident.
EvaluatingValidity
At the end of our journey, Gary Klein and I agreed on a general answer to our initial
question: When can you trust an experienced professional who claims to have an
intuition?Ourconclusionwasthatforthemostpartitispossibletodistinguishintuitions
thatarelikelytobevalidfromthosethatarelikelytobebogus.Asinthejudgmentof
whetheraworkofartisgenuineorafake,youwillusuallydobetterbyfocusingonits
provenancethanbylookingatthepieceitself.Iftheenvironmentissufficientlyregular
andifthejudgehashadachancetolearnitsregularities,theassociativemachinerywill
recognizesituationsandgeneratequickandaccuratepredictionsanddecisions.Youcan
trustsomeone’sintuitionsiftheseconditionsaremet.
Unfortunately, associativentu memory also generates subjectively compelling
intuitions that are false. Anyone who has watched the chess progress of a talented
youngsterknowswellthatskilldoesnotbecomeperfectallatonce,andthatonthewayto
nearperfectionsome mistakesaremade with greatconfidence.When evaluatingexpert
intuitionyoushouldalwaysconsiderwhethertherewasanadequateopportunitytolearn
thecues,eveninaregularenvironment.
Inalessregular,orlow-validity,environment,theheuristicsofjudgmentareinvoked.
System 1 is often able to produce quick answers to difficult questions by substitution,
creatingcoherencewherethereisnone.Thequestionthatisansweredisnottheonethat
wasintended,buttheanswerisproducedquicklyandmaybesufficientlyplausibletopass
thelaxandlenientreviewofSystem2.Youmaywanttoforecastthecommercialfutureof
acompany,forexample,andbelievethatthisiswhatyouarejudging,whileinfactyour
evaluationisdominatedbyyourimpressionsoftheenergyandcompetenceofitscurrent
executives.Becausesubstitutionoccursautomatically,youoftendonotknowtheoriginof
ajudgmentthatyou(yourSystem2)endorseandadopt.Ifitistheonlyonethatcomesto
mind,itmaybesubjectivelyundistinguishablefromvalidjudgmentsthatyoumakewith
expertconfidence.Thisiswhysubjectiveconfidenceisnotagooddiagnosticofaccuracy:
judgmentsthatanswerthewrongquestioncanalsobemadewithhighconfidence.
Youmaybeasking,Whydidn’tGaryKleinandIcomeupimmediatelywiththeidea
ofevaluatinganexpert’sintuitionbyassessingtheregularityoftheenvironmentandthe
expert’slearninghistory—mostlysettingasidetheexpert’sconfidence?Andwhatdidwe
thinktheanswercouldbe?Thesearegoodquestionsbecausethecontoursofthesolution
wereapparentfromthebeginning.Weknewattheoutsetthatfiregroundcommandersand
pediatricnurseswouldendupononesideoftheboundaryofvalidintuitionsandthatthe
specialtiesstudiedbyMeehlwouldbeontheother,alongwithstockpickersandpundits.
Itisdifficulttoreconstructwhatitwasthattookusyears,longhoursofdiscussion,
endlessexchangesofdrafts and hundreds of e-mails negotiatingoverwords,andmore
than once almost giving up. But this is what always happens when a project ends
reasonably well: once you understand the main conclusion, it seems it was always
obvious.
Asthetitleofourarticlesuggests,KleinandIdisagreedlessthanwehadexpected
andacceptedjointsolutionsofalmostallthesubstantiveissuesthatwereraised.However,
wealsofoundthatourearlydifferencesweremorethananintellectualdisagreement.We
haddifferentattitudes,emotions,andtastes,andthosechangedremarkablylittleoverthe
years.Thisismostobviousinthefactsthatwefindamusingandinteresting.Kleinstill
winceswhenthewordbiasismentioned,andhestillenjoysstoriesinwhichalgorithmsor
formal procedures lead to obviously absurd decisions. I tend to view the occasional
failuresofalgorithmsasopportunitiestoimprovethem.Ontheotherhand,Ifindmore
pleasure than Klein does in the come-uppance of arrogant experts who claim intuitive
powersinzero-validitysituations.Inthelongrun,however,findingasmuchintellectual
agreement as we did is surely more important than the persistent emotional differences
thatremained.
SpeakingofExpertIntuition
“Howmuchexpertisedoesshehaveinthisparticulartask?Howmuchpracticehas
shehad?”
“Does he really believe that the environment of start-ups is sufficiently regular to
justifyanintuitionthatgoesagainstthebaserates?”
“Sheisveryconfidentinherdecision,butsubjectiveconfidenceisapoorindexof
theaccuracyofajudgment.”
“Did he really have an opportunity to learn? How quick and how clear was the
feedbackhereceivedonhisjudgments?”
P
TheOutsideView
AfewyearsaftermycollaborationwithAmosbegan,Iconvincedsomeofficialsinthe
IsraeliMinistryofEducationoftheneedforacurriculumtoteachjudgmentanddecision
makinginhighschools.TheteamthatIassembledtodesignthecurriculumandwritea
textbookforit includedseveralexperienced teachers, someof my psychologystudents,
andSeymourFox,thendeanoftheHebrewUniversity’sSchoolofEducation,whowasan
expertincurriculumdevelopment.
AftermeetingeveryFridayafternoonforaboutayear,wehadconstructedadetailed
outlineofthesyllabus,hadwrittenacoupleofchapters,andhadrunafewsamplelessons
in the classroom. We all felt that we had made good progress. One day, as we were
discussing procedures for estimating uncertain quantities, the idea of conducting an
exerciseoccurredtome.Iaskedeveryonetowritedownanestimateofhowlongitwould
take us to submit a finished draft of the textbook to the Ministry of Education. I was
following a procedure that we already planned to incorporate into our curriculum: the
properwaytoelicitinformationfromagroupisnotbystartingwithapublicdiscussion
butbyconfidentiallycollectingeachperson’sjudgment.Thisproceduremakesbetteruse
oftheknowledgeavailabletomembersofthegroupthanthecommonpracticeofopen
discussion.Icollectedtheestimatesandjottedtheresultsontheblackboard.Theywere
narrowlycenteredaroundtwoyears;thelowendwasoneandahalf,thehighendtwoand
ahalfyears.
Then I had another idea. I turned to Seymour, our curriculum expert, and asked
whether he could think of other teams similar to ours that had developed a curriculum
fromscratch.Thiswasatimewhenseveralpedagogicalinnovationslike“newmath”had
beenintroduced,andSeymoursaidhecouldthinkofquiteafew.Ithenaskedwhetherhe
knewthehistoryoftheseteamsinsomedetail,anditturnedoutthathewasfamiliarwith
several.Iaskedhimtothinkoftheseteamswhentheyhadmadeasmuchprogressaswe
had.Howlong,fromthatpoint,didittakethemtofinishtheirtextbookprojects?
He fell silent. When he finally spoke, it seemed to me that he was blushing,
embarrassedbyhisownanswer:“Youknow,Ineverrealizedthisbefore,butinfactnotall
theteamsatastagecomparabletoourseverdidcompletetheirtask.Asubstantialfraction
oftheteamsendedupfailingtofinishthejob.”
Thiswasworrisome;wehadneverconsideredthepossibilitythatwemightfail.My
anxiety rising, I asked how large he estimated that fraction was. Rw l剢 sidering
t20;About40%,”heanswered.Bynow,apallofgloomwasfallingovertheroom.The
nextquestionwasobvious:“Thosewhofinished,”Iasked.“Howlongdidittakethem?”
“Icannotthinkofanygroupthatfinishedinlessthansevenyears,”hereplied,“norany
thattookmorethanten.”
Igraspedatastraw:“Whenyoucompareourskillsandresourcestothoseoftheother
groups, how good are we? How would you rank us in comparison with these teams?”
Seymour did not hesitate long this time. “We’re below average,” he said, “but not by
much.”Thiscame asacomplete surprisetoall ofus—includingSeymour, whose prior
estimate had been well within the optimistic consensus of the group. Until I prompted
him,therewasnoconnectioninhismindbetweenhisknowledgeofthehistoryofother
teamsandhisforecastofourfuture.
OurstateofmindwhenweheardSeymourisnotwelldescribedbystatingwhatwe
“knew.” Surely all of us “knew” that a minimum of seven years and a 40% chance of
failurewasamoreplausibleforecastofthefateofourprojectthanthenumberswehad
writtenonourslipsofpaperafewminutesearlier.Butwedidnotacknowledgewhatwe
knew.Thenewforecaststillseemedunreal,becausewecouldnotimaginehowitcould
takesolongtofinishaprojectthatlookedsomanageable.Nocrystalballwasavailableto
tellusthestrangesequenceofunlikelyeventsthatwereinourfuture.Allwecouldsee
was a reasonable plan that should produce a book in about two years, conflicting with
statistics indicating that other teams had failed or had taken an absurdly long time to
complete their mission. What we had heard was base-rate information, from which we
shouldhaveinferredacausalstory:ifsomanyteamsfailed,andifthosethatsucceeded
tooksolong,writingacurriculumwassurelymuchharderthanwehadthought.Butsuch
aninferencewouldhaveconflictedwithourdirectexperienceofthegoodprogresswehad
beenmaking.ThestatisticsthatSeymourprovidedweretreatedasbaseratesnormallyare
—notedandpromptlysetaside.
Weshouldhavequitthatday.Noneofuswaswillingtoinvestsixmoreyearsofwork
inaprojectwitha40%chanceoffailure.Althoughwemusthavesensedthatpersevering
wasnotreasonable,thewarningdidnotprovideanimmediatelycompellingreasontoquit.
Afterafewminutesofdesultorydebate,wegatheredourselvestogetherandcarriedonas
ifnothinghadhappened.Thebookwaseventuallycompletedeight(!)yearslater.Bythat
time I was no longer living in Israel and had long since ceased to be partof the team,
whichcompletedthetaskaftermanyunpredictablevicissitudes.Theinitialenthusiasmfor
theideaintheMinistryofEducationhadwanedbythetimethetextwasdeliveredandit
wasneverused.
This embarrassing episode remains one of the most instructive experiences of my
professional life. I eventually learned three lessons from it. The first was immediately
apparent:Ihadstumbledontoadistinctionbetweentwoprofoundlydifferentapproaches
toforecasting,whichAmosandIlaterlabeledtheinsideviewandtheoutsideview.The
secondlessonwasthatourinitialforecastsofabouttwoyearsforthecompletionofthe
projectexhibitedaplanningfallacy.Ourestimateswereclosertoabest-casescenariothan
toa realisticassessment. Iwas slowerto acceptthe thirdlesson, whichI callirrational
perseverance:thefollywedisplayedthatdayinfailingtoabandontheproject.Facinga
choice,wegaveuprationalityratherthangiveuptheenterprise.
DrawntotheInsideView
On that long-ago Friday, our curriculum expert made two judgments about the same
problemandarrivedatverydifferentanswers.Theinsideviewistheone thatallofus,
includingSeymour,spontaneouslyadoptedtoassessthefutureofourproject.Wefocused
onourspecificcircumstancesandsearchedforevidenceinourownexperiences.Wehada
sketchyplan:weknewhowmanychaptersweweregoingtowrite,andwehadanideaof
howlongithadtakenustowritethetwothatwehadalreadydone.Themorecautious
amongusprobablyaddedafewmonthstotheirestimateasamarginoferror.
Extrapolatingwasamistake.Wewereforecastingbasedontheinformationinfrontof
us—WYSIATI—butthechapterswewrotefirstwereprobablyeasierthanothers,andour
commitmenttotheprojectwasprobablythenatitspeak.Butthemainproblemwasthat
wefailedtoallowforwhatDonaldRumsfeldfamouslycalledthe“unknownunknowns.”
Therewasnowayforustoforesee,thatday,thesuccessionofeventsthatwouldcausethe
projecttodragoutforsolong.Thedivorces,theillnesses,thecrisesofcoordinationwith
bureaucraciesthatdelayedtheworkcouldnotbeanticipated.Sucheventsnotonlycause
thewritingofchapterstoslowdown,theyalsoproducelongperiodsduringwhichlittleor
noprogressismadeatall.Thesamemusthavebeentrue,ofcourse,fortheotherteams
thatSeymourknewabout.Themembersofthoseteamswerealsounabletoimaginethe
eventsthatwouldcausethemtospendsevenyearstofinish,orultimatelyfailtofinish,a
projectthattheyevidentlyhadthoughtwasveryfeasible.Likeus,theydidnotknowthe
oddstheywerefacing.Therearemanywaysforanyplantofail,andalthoughmostof
themaretooimprobabletobeanticipated,thelikelihoodthatsomethingwillgowrongin
abigprojectishigh.
ThesecondquestionIaskedSeymourdirectedhisattentionawayfromusandtoward
aclassofsimilarcases.Seymourestimatedthebaserateofsuccessinthatreferenceclass:
40%failureandseventotenyearsforcompletion.Hisinformalsurveywassurelynotup
to scientific standards of evidence, but it provided a reasonable basis for a baseline
prediction:thepredictionyoumakeaboutacaseifyouknownothingexceptthecategory
towhichitbelongs.Aswesawearlier,thebaselinepredictionshouldbetheanchorfor
further adjustments. If you are asked to guess the height of a woman about whom you
knowonlythatshelivesinNewYorkCity,yourbaselinepredictionisyourbestguessof
theaverageheightofwomeninthecity.Ifyouarenowgivencase-specificinformation,
forexamplethatthewoman’ssonisthestartingcenterofhishighschoolbasketballteam,
youwilladjustyourestimateawayfromthemeanintheappropriatedirection.Seymour’s
comparisonofourteamtootherssuggestedthattheforecastofouroutcomewasslightly
worsethanthebaselineprediction,whichwasalreadygrim.
The spectacular accuracy of the outside-view forecast in our problem was surely a
flukeandshouldnotcountasevidenceforthevalidityoftheoutsideview.Theargument
fortheoutsideviewshouldbemadeongeneralgrounds:ifthereferenceclassisproperly
chosen, the outside view will give an indication of where the ballpark is, and it may
suggest,asitdidinourcase,thattheinside-viewforecastsarenotevenclosetoit.
Forapsychologist,thediscrepancybetweenSeymour’stwojudgmentsisstriking.He
had in his head all the knowledge required to estimate the statistics of an appropriate
reference class, but he reached his initial estimate without ever using that knowledge.
Seymour’s forecast from his insidethaa view was not an adjustment from the baseline
prediction,whichhadnotcometohismind.Itwasbasedontheparticularcircumstances
ofourefforts.LiketheparticipantsintheTomWexperiment,Seymourknewtherelevant
baseratebutdidnotthinkofapplyingit.
UnlikeSeymour,therestofusdidnothaveaccesstotheoutsideviewandcouldnot
haveproducedareasonablebaselineprediction.Itisnoteworthy,however,thatwedidnot
feelwe neededinformationabout otherteams tomake our guesses.My requestfor the
outsideviewsurprisedallofus,includingme!Thisisacommonpattern:peoplewhohave
informationaboutanindividualcaserarelyfeeltheneedtoknowthestatisticsoftheclass
towhichthecasebelongs.
Whenwewereeventuallyexposedtotheoutsideview,wecollectivelyignoredit.We
can recognize what happened to us; it is similar to the experiment that suggested the
futilityofteachingpsychology.Whentheymadepredictionsaboutindividualcasesabout
whichtheyhad alittleinformation (a briefand blandinterview),Nisbett and Borgida’s
studentscompletelyneglectedtheglobalresultstheyhadjustlearned.“Pallid”statistical
informationisroutinelydiscardedwhenitisincompatiblewithone’spersonalimpressions
ofacase.Inthecompetitionwiththeinsideview,theoutsideviewdoesn’tstandachance.
Thepreferencefortheinsideviewsometimescarriesmoralovertones.Ionceasked
my cousin, a distinguished lawyer, a question about a reference class: “What is the
probabilityofthedefendantwinningincaseslikethisone?”Hissharpanswerthat“every
case is unique” was accompanied by a look that made it clear he found my question
inappropriate and superficial. A proud emphasis on the uniqueness of cases is also
commoninmedicine,inspiteofrecentadvancesinevidence-basedmedicinethatpoint
the other way. Medical statistics and baseline predictions come up with increasing
frequency in conversations between patients and physicians. However, the remaining
ambivalence about the outside view in the medical profession is expressed in concerns
abouttheimpersonalityofproceduresthatareguidedbystatisticsandchecklists.
ThePlanningFallacy
Inlightofboththeoutside-viewforecastandtheeventualoutcome,theoriginalestimates
we made that Friday afternoon appear almost delusional. This should not come as a
surprise: overly optimistic forecasts of the outcome of projects are found everywhere.
AmosandIcoinedthetermplanningfallacytodescribeplansandforecaststhat
areunrealisticallyclosetobest-casescenarios
couldbeimprovedbyconsultingthestatisticsofsimilarcases
Examples of the planning fallacy abound in the experiences of individuals,
governments,andbusinesses.Thelistofhorrorstoriesisendless.
In July 1997, the proposed new Scottish Parliament building in Edinburgh was
estimatedtocostupto£40million.ByJune1999,thebudgetforthebuildingwas
£109million.InApril2000,legislatorsimposeda£195million“caponcosts.”By
November2001,theydemandedanestimateof“finalcost,”whichwassetat£241
million. That estimated final cost rose twice in 2002, ending the year at £294.6
million. It rose three times more in 2003, reaching £375.8 million by June. The
building was finally comanspleted in 2004 at an ultimate cost of roughly £431
million.
A2005studyexaminedrailprojectsundertakenworldwidebetween1969and1998.
Inmorethan90%ofthecases,thenumberofpassengersprojectedtousethesystem
wasoverestimated.Eventhoughthesepassengershortfallswerewidelypublicized,
forecastsdidnotimproveoverthosethirtyyears;onaverage,plannersoverestimated
how many people would use the new rail projects by 106%, and the average cost
overrunwas45%.Asmoreevidenceaccumulated,theexpertsdidnotbecomemore
reliantonit.
In2002,asurveyofAmericanhomeownerswhohadremodeledtheirkitchensfound
that,on average,theyhadexpectedthejobto cost$18,658;infact,theyendedup
payinganaverageof$38,769.
The optimism of planners and decision makers is not the only cause of overruns.
Contractorsofkitchenrenovations and ofweaponsystems readily admit(thoughnotto
theirclients)thattheyroutinelymakemostoftheirprofitonadditionstotheoriginalplan.
Thefailuresofforecastinginthesecasesreflectthecustomers’inabilitytoimaginehow
much their wishes will escalate over time. They end up paying much more than they
wouldiftheyhadmadearealisticplanandstucktoit.
Errorsintheinitialbudgetarenotalwaysinnocent.Theauthorsofunrealisticplans
areoftendrivenbythedesiretogettheplanapproved—whetherbytheirsuperiorsorbya
client—supportedbytheknowledgethatprojectsarerarelyabandonedunfinishedmerely
becauseofoverrunsincostsorcompletiontimes.Insuchcases,thegreatestresponsibility
foravoidingtheplanningfallacylieswiththedecisionmakerswhoapprovetheplan.If
theydonotrecognizetheneedforanoutsideview,theycommitaplanningfallacy.
MitigatingthePlanningFallacy
The diagnosis of and the remedy for the planning fallacy have not changed since that
Fridayafternoon,buttheimplementationoftheideahascomealongway.Therenowned
Danish planning expert Bent Flyvbjerg, now at Oxford University, offered a forceful
summary:
The prevalent tendency to underweight or ignore distributional information is
perhaps the major source of error in forecasting. Planners should therefore make
every effort to frame the forecasting problem so as to facilitate utilizing all the
distributionalinformationthatisavailable.
This may be considered the single most important piece of advice regarding how to
increase accuracy in forecasting through improved methods. Using such distributional
information from other ventures similar to that being forecasted is called taking an
“outsideview”andisthecuretotheplanningfallacy.
Thetreatmentfortheplanningfallacyhasnowacquiredatechnicalname,reference
class forecasting, and Flyvbjerg has applied it to transportation projects in several
countries. The outside view is implemented by using a large database, which provides
informationonbothplansandoutcomesforhundredsofprojectsallovertheworld,and
canbeusedtoprovidestatisticalinformationaboutthelikelyoverrunsofcostandtime,
andaboutthelikelyunderperformanceofprojectsofdifferenttypes.
The forecasting method that Flyvbjerg applies is similar to the practices
recommendedforovercomingbase-rateneglect:
1. Identify an appropriate reference class (kitchen renovations, large railway projects,
etc.).
2. Obtainthestatisticsofthereferenceclass(intermsofcostpermileofrailway,orof
thepercentagebywhichexpendituresexceededbudget).Usethestatisticstogenerate
abaselineprediction.
3. Usespecificinformationaboutthecasetoadjustthebaselineprediction,ifthereare
particularreasonstoexpecttheoptimisticbiastobemoreorlesspronouncedinthis
projectthaninothersofthesametype.
Flyvbjerg’sanalysesareintendedtoguidetheauthoritiesthatcommissionpublicprojects,
byprovidingthestatisticsofoverrunsinsimilarprojects.Decisionmakersneedarealistic
assessment of the costs and benefits of a proposal before making the final decision to
approve it. They may also wish to estimate the budget reserve that they need in
anticipation of overruns, although such precautions often become self-fulfilling
prophecies.AsoneofficialtoldFlyvbjerg,“Abudgetreserveistocontractorsasredmeat
istolions,andtheywilldevourit.”
Organizationsfacethechallengeofcontrollingthetendencyofexecutivescompeting
for resources to present overly optimistic plans. A well-run organization will reward
plannersforpreciseexecutionandpenalizethemforfailingtoanticipatedifficulties,and
for failing to allow for difficulties that they could not have anticipated—the unknown
unknowns.
DecisionsandErrors
ThatFridayafternoonoccurredmorethanthirtyyearsago.I often thought aboutitand
mentioneditinlecturesseveraltimeseachyear.Someofmyfriendsgotboredwiththe
story,butIkeptdrawingnewlessonsfromit.AlmostfifteenyearsafterIfirstreportedon
theplanning fallacywith Amos, Ireturned to the topic withDan Lovallo.Togetherwe
sketchedatheoryofdecisionmakinginwhichtheoptimisticbiasisasignificantsourceof
risk taking. In the standard rational model of economics, people take risks because the
odds are favorable—they accept some probability of a costly failure because the
probabilityofsuccessissufficient.Weproposedanalternativeidea.
Whenforecastingtheoutcomesofriskyprojects,executivestooeasilyfallvictimto
theplanningfallacy.Initsgrip,theymakedecisionsbasedondelusionaloptimismrather
thanonarationalweightingofgains,losses,andprobabilities.Theyoverestimatebenefits
andunderestimatecosts.Theyspinscenariosofsuccesswhileoverlookingthepotential
formistakesandmiscalculations.Asaresult,theypursueinitiativesthatareunlikelyto
comeinonbudgetorontimeortodelivertheexpectedreturns—oreventobecompleted.
In this view, people often (but not always) take on risky projects because they are
overlyoptimisticabouttheoddstheyface.Iwillreturntothisideaseveraltimesinthis
book—it probably contributes to an explanation of why people litigate, why they start
wars,andwhytheyopensmallbusinesses.
FailingaTest
For many years, I thought that the main point of the curriculum story was what I had
learnedaboutmyfriendSeymour:thathisbestguessaboutthefutureofourprojectwas
notinformedbywhatheknewaboutsimilarprojects.Icameoffquitewellinmytelling
ofthestory,irInwhichIhadtheroleofcleverquestionerandastutepsychologist.Ionly
recentlyrealizedthatIhadactuallyplayedtherolesofchiefdunceandineptleader.
Theprojectwasmyinitiative,anditwasthereforemyresponsibilitytoensurethatit
madesenseandthatmajorproblemswereproperlydiscussedbytheteam,butIfailedthat
test.Myproblemwasnolongertheplanningfallacy.Iwascuredofthatfallacyassoonas
I heard Seymour’s statistical summary. If pressed, I would have said that our earlier
estimateshadbeenabsurdlyoptimistic.Ifpressedfurther,Iwouldhaveadmittedthatwe
hadstartedtheprojectonfaultypremisesandthatweshouldatleastconsiderseriouslythe
option of declaring defeat and going home. But nobody pressed me and there was no
discussion;wetacitlyagreedtogoonwithoutanexplicitforecastofhowlongtheeffort
wouldlast.Thiswaseasytodobecausewehadnotmadesuchaforecasttobeginwith.If
wehadhad areasonablebaselineprediction when westarted,we would nothavegone
intoit,butwe had alreadyinvestedagreatdeal of effort—aninstanceofthe sunk-cost
fallacy,which wewilllookat more closelyinthenext part ofthebook. It wouldhave
beenembarrassingforus—especiallyforme—togiveupatthatpoint,andthereseemed
tobenoimmediatereasontodoso.Itiseasiertochangedirectionsinacrisis,butthiswas
notacrisis,onlysomenewfactsabout people we did not know.Theoutsideviewwas
mucheasiertoignorethanbadnewsinourowneffort.Icanbestdescribeourstateasa
formoflethargy—anunwillingnesstothinkaboutwhathadhappened.Sowecarriedon.
There was no further attempt at rational planning for the rest of the time I spent as a
memberoftheteam—aparticularlytroublingomissionforateamdedicatedtoteaching
rationality.IhopeIamwisertoday,andIhaveacquiredahabitoflookingfortheoutside
view.Butitwillneverbethenaturalthingtodo.
SpeakingoftheOutsideView
“He’stakinganinsideview.Heshouldforgetabouthisowncaseandlookforwhat
happenedinothercases.”
“She is the victim of a planning fallacy. She’s assuming a best-case scenario, but
therearetoomanydifferentwaysfortheplantofail,andshecannotforeseethem
all.”
“Suppose you did not know a thing about this particular legal case, only that it
involvesamalpracticeclaimbyanindividualagainstasurgeon.Whatwouldbeyour
baselineprediction?Howmanyofthesecasessucceedincourt?Howmanysettle?
Whataretheamounts?Isthecasewearediscussingstrongerorweakerthansimilar
claims?”
“Wearemakinganadditionalinvestmentbecausewedonotwanttoadmitfailure.
Thisisaninstanceofthesunk-costfallacy.”
P
TheEngineofCapitalism
Theplanningfallacyisonlyoneofthemanifestationsofapervasiveoptimisticbias.sidto
adtionsofaMostofusviewtheworldasmorebenignthanitreallyis,ourownattributes
asmorefavorablethantheytrulyare,andthegoalsweadoptasmoreachievablethanthey
arelikelytobe.Wealsotendtoexaggerateourabilitytoforecastthefuture,whichfosters
optimisticoverconfidence.Intermsofitsconsequencesfordecisions,theoptimisticbias
maywellbethemostsignificantofthecognitivebiases.Becauseoptimisticbiascanbe
bothablessingandarisk,youshouldbebothhappyandwaryifyouaretemperamentally
optimistic.
Optimists
Optimismisnormal,butsomefortunatepeoplearemoreoptimisticthantherestofus.If
youaregeneticallyendowedwithanoptimisticbias,youhardlyneedtobetoldthatyou
arealuckyperson—youalreadyfeelfortunate.Anoptimisticattitudeislargelyinherited,
anditispartofageneraldispositionforwell-being,whichmayalsoincludeapreference
for seeing the bright side of everything. If you were allowed one wish for your child,
seriously consider wishing him or her optimism. Optimists are normally cheerful and
happy,andthereforepopular;theyareresilientinadaptingtofailuresandhardships,their
chances of clinical depression are reduced, their immune system is stronger, they take
better care of their health, they feel healthier than others and are in fact likely to live
longer. A study of people who exaggerate their expected life span beyond actuarial
predictions showed that they work longer hours, are more optimistic about their future
income, are more likely to remarry after divorce (the classic “triumph of hope over
experience”),andaremorepronetobetonindividualstocks.Ofcourse,theblessingsof
optimismareofferedonlytoindividualswhoareonlymildlybiasedandwhoareableto
“accentuatethepositive”withoutlosingtrackofreality.
Optimistic individuals play a disproportionate role in shaping our lives. Their
decisions make a difference; they are the inventors, the entrepreneurs, the political and
militaryleaders—notaveragepeople.Theygottowheretheyarebyseekingchallenges
andtakingrisks.Theyaretalentedandtheyhavebeenlucky,almostcertainlyluckierthan
theyacknowledge.Theyareprobablyoptimisticbytemperament;asurveyoffoundersof
smallbusinessesconcludedthatentrepreneursaremoresanguinethanmidlevelmanagers
about life in general. Their experiences of success have confirmed their faith in their
judgmentandintheirabilitytocontrolevents.Theirself-confidenceisreinforcedbythe
admiration of others. This reasoning leads to a hypothesis: the people who have the
greatestinfluenceonthelivesofothersarelikelytobeoptimisticandoverconfident,and
totakemorerisksthantheyrealize.
Theevidencesuggeststhatanoptimisticbiasplaysarole—sometimesthedominantrole
—whenever individuals or institutions voluntarily take on significant risks. More often
thannot, risktakers underestimate theodds theyface, and do invest sufficienteffortto
findoutwhattheoddsare.Becausetheymisreadtherisks,optimisticentrepreneursoften
believetheyareprudent,evenwhentheyarenot.Theirconfidenceintheirfuturesuccess
sustainsapositivemoodthathelpsthemobtainresourcesfromothers,raisethemoraleof
their employees, and enhance their prospects of prevailing. When action is needed,
optimism,evenofthemildlydelusionalvariety,maybeagoodthing.
EntrepreneurialDelusions
ThechancesthatasmallbusinesswillthesurviveforfiveyearsintheUnitedStatesare
about35%.Buttheindividualswhoopensuchbusinessesdonotbelievethatthestatistics
applytothem.AsurveyfoundthatAmericanentrepreneurstendtobelievetheyareina
promising line of business: their average estimate of the chances of success for “any
businesslikeyours”was60%—almostdoublethetruevalue.Thebiaswasmoreglaring
whenpeopleassessedtheoddsoftheirownventure.Fully81%oftheentrepreneursput
theirpersonaloddsofsuccessat7outof10orhigher,and33%saidtheirchanceoffailing
waszero.
Thedirectionofthebiasisnotsurprising.Ifyouinterviewedsomeonewhorecently
opened an Italian restaurant, you would not expect her to have underestimated her
prospectsforsuccessortohaveapoorviewofherabilityasarestaurateur.Butyoumust
wonder:Wouldshestillhaveinvestedmoneyandtimeifshehadmadeareasonableeffort
tolearntheodds—or,ifshedidlearntheodds(60%ofnewrestaurantsareoutofbusiness
afterthreeyears),paidattentiontothem?Theideaofadoptingtheoutsideviewprobably
didn’toccurtoher.
Oneofthebenefitsofanoptimistictemperamentisthatitencouragespersistencein
the face of obstacles. But persistence can be costly. An impressive series of studies by
ThomasÅstebroshedslightonwhathappenswhenoptimistsreceivebadnews.Hedrew
his data from a Canadian organization—the Inventor’s Assistance Program—which
collectsasmallfeetoprovideinventorswithanobjectiveassessmentofthecommercial
prospects of their idea. The evaluations rely on careful ratings of each invention on 37
criteria,includingneedfortheproduct,costofproduction,andestimatedtrendofdemand.
Theanalystssummarizetheirratingsbyalettergrade,whereDandEpredictfailure—a
predictionmadeforover70%oftheinventionstheyreview.Theforecastsoffailureare
remarkably accurate: only 5 of 411 projects that were given the lowest grade reached
commercialization,andnonewassuccessful.
Discouragingnewsledabouthalfoftheinventorstoquitafterreceivingagradethat
unequivocally predicted failure. However, 47% of them continued development efforts
evenafterbeingtoldthattheirprojectwashopeless,andonaveragethesepersistent(or
obstinate) individuals doubled their initial losses before giving up. Significantly,
persistenceafterdiscouragingadvicewasrelativelycommonamonginventorswhohada
highscoreon apersonalitymeasure ofoptimism—onwhich inventors generallyscored
higher than the general population. Overall, the return on private invention was small,
“lowerthanthereturnonprivateequityandonhigh-risksecurities.”Moregenerally,the
financialbenefitsofself-employmentaremediocre:giventhesamequalifications,people
achievehigheraveragereturnsbysellingtheirskillstoemployersthanbysettingouton
theirown.Theevidencesuggeststhatoptimismiswidespread,stubborn,andcostly.
Psychologists have confirmed that most people genuinely believe that they are
superiortomostothersonmostdesirabletraits—theyarewillingtobetsmallamountsof
money on these beliefs in the laboratory. In the market, of course, beliefs in one’s
superiority have significant consequences. Leaders of large businesses sometimes make
hugebetsinexpensivemergersandacquisitions,actingonthemistakenbeliefthatthey
canmanagethe assets ofanothercompany better thanitscurrent owners do.Thestock
market commonly responds by downgrading the value of the acquiring firm, because
experience has shown that efforts to integrate large firms fail more often than they
succeed.The misguidedacquisitions havebeen explainedby a “hubris hypothesis”: the
eivxecutivesoftheacquiringfirmaresimplylesscompetentthantheythinktheyare.
TheeconomistsUlrikeMalmendierandGeoffreyTateidentifiedoptimisticCEOsby
the amount of company stock that they owned personally and observed that highly
optimistic leaders took excessive risks. They assumed debt rather than issue equity and
were more likely than others to “overpay for target companies and undertake value-
destroying mergers.” Remarkably, the stock of the acquiring company suffered
substantiallymoreinmergersiftheCEOwasoverlyoptimisticbytheauthors’measure.
The stock market is apparently able to identify overconfident CEOs. This observation
exoneratestheCEOsfromoneaccusationevenasitconvictsthemofanother:theleaders
ofenterpriseswhomakeunsoundbetsdonotdosobecausetheyarebettingwithother
people’smoney.Onthecontrary,theytakegreaterriskswhentheypersonallyhavemore
atstake.ThedamagecausedbyoverconfidentCEOsiscompoundedwhenthebusiness
pressanointsthemascelebrities;theevidenceindicatesthatprestigiouspressawardsto
theCEOarecostlytostockholders.Theauthorswrite,“We findthatfirmswithaward-
winning CEOs subsequently underperform, in terms both of stock and of operating
performance.Atthesametime,CEOcompensationincreases,CEOsspendmoretimeon
activitiesoutsidethecompanysuchas writing booksandsittingonoutsideboards, and
theyaremorelikelytoengageinearningsmanagement.”
Manyyearsago,mywifeandIwereonvacationonVancouverIsland,lookingforaplace
tostay.Wefoundanattractivebutdesertedmotelonalittle-traveledroadinthemiddleof
aforest.Theownerswereacharmingyoungcouplewhoneededlittlepromptingtotellus
theirstory.TheyhadbeenschoolteachersintheprovinceofAlberta;theyhaddecidedto
change their life and used their life savings to buy this motel, which had been built a
dozenyearsearlier.Theytolduswithoutironyorself-consciousnessthattheyhadbeen
abletobuyitcheap,“becausesixorsevenpreviousownershadfailedtomakeagoofit.”
Theyalsotoldusaboutplanstoseekaloantomaketheestablishmentmoreattractiveby
buildingarestaurantnexttoit.Theyfeltnoneedtoexplainwhytheyexpectedtosucceed
wheresixorsevenothershadfailed.Acommonthreadofboldnessandoptimismlinks
businesspeople,frommotelownerstosuperstarCEOs.
The optimistic risk taking of entrepreneurs surely contributes to the economic
dynamismofacapitalisticsociety,evenifmostrisktakersendupdisappointed.However,
Marta Coelho of the London School of Economics has pointed out the difficult policy
issuesthatarisewhenfoundersofsmallbusinessesaskthegovernmenttosupportthemin
decisions that are most likely to end badly. Should the government provide loans to
would-be entrepreneurs who probably will bankrupt themselves in a few years? Many
behavioraleconomistsarecomfortablewiththe“libertarianpaternalistic”proceduresthat
help people increase their savings rate beyond what they would do on their own. The
questionofwhetherandhowgovernmentshouldsupportsmallbusinessdoesnothavean
equallysatisfyinganswer.
CompetitionNeglect
Itistemptingtoexplainentrepreneurialoptimismbywishfulthinking,butemotionisonly
part of the story. Cognitive biases play an important role, notably the System 1 feature
WYSIATI.
Wefocusonourgoal,anchoronourplan,andneglectrelevantbaserates,exposing
ourselvestotneseheplanningfallacy.
Wefocusonwhatwewanttodoandcando,neglectingtheplansandskillsofothers.
Bothinexplainingthepastandinpredictingthefuture,wefocusonthecausalroleof
skillandneglecttheroleofluck.Wearethereforepronetoanillusionofcontrol.
We focus on what we know and neglect what we do not know, which makes us
overlyconfidentinourbeliefs.
Theobservationthat“90%ofdriversbelievetheyarebetterthanaverage”isawell-
establishedpsychologicalfindingthathasbecomepartoftheculture,anditoftencomes
upasaprimeexampleofamoregeneralabove-averageeffect.However,theinterpretation
ofthefindinghaschangedinrecentyears,fromself-aggrandizementtoacognitivebias.
Considerthesetwoquestions:
Areyouagooddriver?
Areyoubetterthanaverageasadriver?
Thefirstquestioniseasyandtheanswercomesquickly:mostdriverssayyes.Thesecond
questionismuchharderandformostrespondentsalmostimpossibletoanswerseriously
andcorrectly,becauseitrequiresanassessmentoftheaveragequalityofdrivers.Atthis
pointin the book it comesas no surprise that peoplerespond toa difficultquestion by
answeringaneasierone.Theycomparethemselvestotheaveragewithouteverthinking
about the average. The evidence for the cognitive interpretation of the above-average
effectisthatwhenpeopleareaskedaboutatasktheyfinddifficult(formanyofusthis
could be “Are you better than average in starting conversations with strangers?”), they
readily rate themselves as below average. The upshot is that people tend to be overly
optimisticabouttheirrelativestandingonanyactivityinwhichtheydomoderatelywell.
Ihavehadseveraloccasionstoaskfoundersandparticipantsininnovativestart-upsa
question:Towhatextentwilltheoutcomeofyoureffortdependonwhatyoudoinyour
firm? This is evidently an easy question; the answer comes quickly and in my small
sampleithasneverbeenlessthan80%.Evenwhentheyarenotsuretheywillsucceed,
theseboldpeoplethinktheirfateisalmostentirelyintheirownhands.Theyaresurely
wrong:theoutcomeofastart-updependsasmuchontheachievementsofitscompetitors
andonchangesinthemarketasonitsownefforts.However,WYSIATIplaysitspart,and
entrepreneurs naturally focus on what they know best—their plans and actions and the
mostimmediatethreatsandopportunities,suchastheavailabilityoffunding.Theyknow
lessabouttheircompetitorsandthereforefinditnaturaltoimagineafutureinwhichthe
competitionplayslittlepart.
Colin Camerer and Dan Lovallo, who coined the concept of competition neglect,
illustrateditwithaquotefromthethenchairmanofDisneyStudios.Askedwhysomany
expensivebig-budgetmoviesarereleasedonthesamedays(suchasMemorialDayand
IndependenceDay),hereplied:
Hubris. Hubris. If you only think about your own business, you think, “I’ve got a
goodstorydepartment,I’vegotagoodmarketingdepartment,we’regoingtogoout
anddothis.”Andyoudon’tthinkthateverybodyelseisthinkingthesameway.Ina
given weekend in a year you’ll have five movies open, and there’s certainly not
enoughpeopletogoaround.re
Thecandidanswerreferstohubris,butitdisplaysnoarrogance,noconceitofsuperiority
to competing studios. The competition is simply not part of the decision, in which a
difficult question has again been replaced by an easier one. The question that needs an
answeristhis:Consideringwhatotherswilldo,howmanypeoplewillseeourfilm?The
questionthestudioexecutivesconsideredissimplerandreferstoknowledgethatismost
easilyavailabletothem:Dowehaveagoodfilmandagoodorganizationtomarketit?
ThefamiliarSystem1processesofWYSIATIandsubstitutionproducebothcompetition
neglectandtheabove-averageeffect.Theconsequenceofcompetitionneglectisexcess
entry:morecompetitorsenterthemarketthanthemarketcanprofitablysustain,sotheir
average outcome is a loss. The outcome is disappointing for the typical entrant in the
market,buttheeffectontheeconomyasawholecouldwellbepositive.Infact,Giovanni
DosiandDanLovallocallentrepreneurialfirmsthatfailbutsignalnewmarketstomore
qualified competitors “optimistic martyrs”—good for the economy but bad for their
investors.
Overconfidence
For a number of years, professors at Duke University conducted a survey in which the
chief financial officers of large corporations estimated the returns of the Standard &
Poor’sindexoverthefollowingyear.TheDukescholarscollected11,600suchforecasts
and examined their accuracy. The conclusion was straightforward: financial officers of
large corporations had no clue about the short-term future of the stock market; the
correlationbetweentheirestimatesandthetruevaluewasslightlylessthanzero!When
theysaidthemarketwouldgodown,itwasslightlymorelikelythannotthatitwouldgo
up.Thesefindingsarenotsurprising.ThetrulybadnewsisthattheCFOsdidnotappear
toknowthattheirforecastswereworthless.
InadditiontotheirbestguessaboutS&Preturns,theparticipantsprovidedtwoother
estimates:avaluethattheywere90%surewouldbetoohigh,andonethattheywere90%
surewouldbetoolow.Therangebetweenthetwovaluesiscalledan“80%confidence
interval”andoutcomesthatfalloutsidetheintervalarelabeled“surprises.”Anindividual
whosetsconfidenceintervalsonmultipleoccasionsexpectsabout20%oftheoutcomesto
besurprises.Asfrequentlyhappensinsuchexercises,therewerefartoomanysurprises;
theirincidencewas67%,morethan3timeshigherthanexpected.ThisshowsthatCFOs
were grossly overconfident about their ability to forecast the market. Overconfidence is
anothermanifestationofWYSIATI:whenweestimateaquantity,werelyoninformation
that comes to mind and construct a coherent story in which the estimate makes sense.
Allowing for the information that does not come to mind—perhaps because one never
knewit—isimpossible.
Theauthorscalculatedtheconfidenceintervalsthatwouldhavereducedtheincidence
of surprises to 20%. The results were striking. To maintain the rate of surprises at the
desiredlevel,theCFOsshouldhavesaid,yearafteryear,“Thereisan80%chancethatthe
S&P return next year will be between –10% and +30%.” The confidence interval that
properly reflects the CFOs’ knowledge (more precisely, their ignorance) is more than 4
timeswiderthantheintervalstheyactuallystated.
Socialpsychologycomesintothepicturehere,becausetheanswerthatatruthfulCFO
wouldofferisplainlyridiculous.ACFOwhoinformshiscolleaguesthat“th%”>iereisa
good chance that the S&P returns will be between –10% and +30%” can expect to be
laughedoutoftheroom.Thewideconfidenceintervalisaconfessionofignorance,which
is not socially acceptable for someone who is paid to be knowledgeable in financial
matters.Eveniftheyknewhowlittletheyknow,theexecutiveswouldbepenalizedfor
admittingit.PresidentTrumanfamouslyaskedfora“one-armedeconomist”whowould
takeaclear stand; hewas sick andtiredof economistswhokeptsaying, “Ontheother
hand…”
Organizations that take the word of overconfident experts can expect costly
consequences. The study of CFOs showed that those who were most confident and
optimisticabouttheS&Pindexwerealsooverconfidentandoptimisticabouttheprospects
of their own firm, which went on to take more risk than others. As Nassim Taleb has
argued, inadequate appreciation of the uncertainty of the environment inevitably leads
economic agents to take risks they should avoid. However, optimism is highly valued,
socially and in the market; people and firms reward the providers of dangerously
misleading information more than they reward truth tellers. One of the lessons of the
financial crisis that led to the Great Recession is that there are periods in which
competition,amongexpertsandamongorganizations,createspowerfulforcesthatfavora
collectiveblindnesstoriskanduncertainty.
The social and economic pressures that favor overconfidence are not restricted to
financialforecasting.Otherprofessionalsmustdealwiththefactthatanexpertworthyof
the name is expected to display high confidence. Philip Tetlock observed that the most
overconfidentexpertswerethemostlikelytobeinvitedtostruttheirstuffinnewsshows.
Overconfidencealsoappearstobeendemicinmedicine.Astudyofpatientswhodiedin
theICUcomparedautopsyresultswiththediagnosisthatphysicianshadprovidedwhile
the patients were still alive. Physicians also reported their confidence. The result:
“clinicianswhowere‘completelycertain’ofthediagnosisantemortemwerewrong40%
ofthetime.”Hereagain,expertoverconfidenceisencouragedbytheirclients:“Generally,
it is considered a weakness and a sign of vulnerability for clinicians to appear unsure.
Confidenceisvaluedoveruncertaintyandthereisaprevailingcensureagainstdisclosing
uncertaintytopatients.”Expertswhoacknowledgethefullextentoftheirignorancemay
expecttobereplacedbymoreconfidentcompetitors,whoarebetterabletogainthetrust
ofclients.Anunbiasedappreciationofuncertaintyisacornerstoneofrationality—butitis
not what people and organizations want. Extreme uncertainty is paralyzing under
dangerous circumstances, and the admission that one is merely guessing is especially
unacceptable when the stakes are high. Acting on pretended knowledge is often the
preferredsolution.
When they come together, the emotional, cognitive, and social factors that support
exaggeratedoptimismareaheadybrew,whichsometimesleadspeopletotakerisksthat
they would avoid if they knew the odds. There is no evidence that risk takers in the
economicdomainhaveanunusualappetiteforgamblesonhighstakes;theyaremerely
lessawareofrisksthanmoretimidpeopleare.DanLovalloandIcoinedthephrase“bold
forecastsandtimiddecisions”todescribethebackgroundofrisktaking.
Theeffectsofhighoptimismondecisionmakingare,atbest,amixedblessing,butthe
contributionofoptimismtogoodimplementationiscertainlypositive.Themainbenefitof
optimismisresilienceinthefaceofsetbacks.AccordingtoMartinSeligman,thefounder
of potelsitive psychology, an “optimistic explanation style” contributes to resilience by
defending one’s self-image. In essence, the optimistic style involves taking credit for
successesbutlittleblameforfailures.Thisstylecanbetaught,atleasttosomeextent,and
Seligman has documented the effects of training on various occupations that are
characterizedbyahighrateoffailures,suchas cold-call sales of insurance(acommon
pursuitinpre-Internetdays).Whenonehasjusthadadoorslammedinone’sfacebyan
angryhomemaker,thethoughtthat“shewasanawfulwoman”isclearlysuperiorto“Iam
an inept salesperson.” I have always believed that scientific research is another domain
whereaformofoptimismisessentialtosuccess:Ihaveyettomeetasuccessfulscientist
who lacks the ability to exaggerate the importance of what he or she is doing, and I
believethatsomeonewholacksadelusionalsenseofsignificancewillwiltinthefaceof
repeated experiences of multiple small failures and rare successes, the fate of most
researchers.
ThePremortem:APartialRemedy
Can overconfident optimism be overcome by training? I am not optimistic. There have
been numerous attempts to train people to state confidence intervals that reflect the
imprecisionoftheirjudgments,withonlyafewreportsofmodestsuccess.Anoftencited
example is that geologists at Royal Dutch Shell became less overconfident in their
assessmentsofpossibledrillingsitesaftertrainingwithmultiplepastcasesforwhichthe
outcome was known. In other situations, overconfidence was mitigated (but not
eliminated) when judges were encouraged to consider competing hypotheses. However,
overconfidenceisadirectconsequenceoffeaturesofSystem1thatcanbetamed—butnot
vanquished. The main obstacle is that subjective confidence is determined by the
coherence of the story one has constructed, not by the quality and amount of the
informationthatsupportsit.
Organizationsmaybebetterabletotameoptimismandindividualsthanindividuals
are. The best idea for doing so was contributed by Gary Klein, my “adversarial
collaborator”whogenerallydefendsintuitivedecisionmakingagainstclaimsofbiasand
istypicallyhostiletoalgorithms.Helabelshisproposalthepremortem.Theprocedureis
simple: when the organization has almost come to an important decision but has not
formally committed itself, Klein proposes gathering for a brief session a group of
individuals who are knowledgeable about the decision. The premise of the session is a
shortspeech:“Imaginethatweareayearintothefuture.Weimplementedtheplanasit
nowexists.Theoutcomewasadisaster.Pleasetake5to10minutestowriteabriefhistory
ofthatdisaster.”
Gary Klein’s idea of the premortem usually evokes immediate enthusiasm. After I
describeditcasuallyatasessioninDavos,someonebehindmemuttered,“Itwasworth
comingtoDavosjustforthis!”(IlaternoticedthatthespeakerwastheCEOofamajor
international corporation.) The premortem has two main advantages: it overcomes the
groupthink that affects many teams once a decision appears to have been made, and it
unleashestheimaginationofknowledgeableindividualsinamuch-neededdirection.
Asateamconvergesonadecision—andespeciallywhentheleadertipsherhand—
public doubts about the wisdom of the planned move are gradually suppressed and
eventuallycometobetreatedasevidenceofflawedloyaltytotheteamanditsleaders.
Thesuppressionofdoubtcontributestooverconfidenceinagroupwhereonlysupporters
ofthedecisionhaveavfilepos-id=“filepos726557”>naceaanddoesnotprovidecomplete
protectionagainstnastysurprises,butitgoessomewaytowardreducingthedamageof
plansthataresubjecttothebiasesofWYSIATIanduncriticaloptimism.
SpeakingofOptimism
“Theyhaveanillusionofcontrol.Theyseriouslyunderestimatetheobstacles.”
“Theyseemtosufferfromanacutecaseofcompetitorneglect.”
“Thisisacaseofoverconfidence.Theyseemtobelievetheyknowmorethanthey
actuallydoknow.”
“Weshouldconductapremortemsession.Someonemaycomeupwithathreatwe
haveneglected.”
P
Part4
P
Choices
P
Bernoulli’sErrors
One day in the early 1970s, Amos handed me a mimeographed essay by a Swiss
economist named Bruno Frey, which discussed the psychological assumptions of
economictheory.Ivividlyrememberthecolorofthecover:darkred.BrunoFreybarely
recallswritingthepiece,butIcanstillreciteitsfirstsentence:“Theagentofeconomic
theoryisrational,selfish,andhistastesdonotchange.”
Iwasastonished.Myeconomistcolleaguesworkedinthebuildingnextdoor,butI
had not appreciated the profound difference between our intellectual worlds. To a
psychologist,itisself-evidentthatpeopleareneitherfullyrationalnorcompletelyselfish,
andthat theirtastesare anythingbut stable.Our two disciplinesseemed tobe studying
differentspecies,whichthebehavioraleconomistRichardThalerlaterdubbedEconsand
Humans.
UnlikeEcons,theHumansthatpsychologistsknowhaveaSystem1.Theirviewof
theworldislimitedbytheinformationthatisavailableatagivenmoment(WYSIATI),
and therefore they cannot be as consistent and logical as Econs. They are sometimes
generousandoftenwillingtocontributetothegrouptowhichtheyareattached.Andthey
often have little idea of what they will like next year or even tomorrow. Here was an
opportunityforaninterestingconversationacrosstheboundariesofthedisciplines.Idid
notanticipatethatmycareerwouldbedefinedbythatconversation.
SoonafterheshowedmeFrey’sarticle,Amossuggestedthatwemakethestudyof
decisionmakingournextproject.Iknewnexttonothingaboutthetopic,butAmoswasan
expertandastarofthefield,andheMathematicalPsychology,andhedirectedmetoa
fewchaptersthathethoughtwouldbeagoodintroduction.
Isoonlearnedthatoursubjectmatterwouldbepeople’sattitudestoriskyoptionsand
that we would seek to answer a specific question: What rules govern people’s choices
betweendifferentsimplegamblesandbetweengamblesandsurethings?
Simple gambles (such as “40% chance to win $300”) are to students of decision
makingwhatthefruitflyistogeneticists.Choicesbetweensuchgamblesprovideasimple
model that shares important features with the more complex decisions that researchers
actuallyaimtounderstand.Gamblesrepresentthefactthattheconsequencesofchoices
arenevercertain.Evenostensiblysureoutcomesareuncertain:whenyousignthecontract
tobuyanapartment,youdonotknowthepriceatwhichyoulatermayhavetosellit,nor
doyouknowthatyourneighbor’ssonwillsoontakeupthetuba.Everysignificantchoice
wemakeinlifecomeswithsomeuncertainty—whichiswhystudentsofdecisionmaking
hopethatsomeofthelessonslearnedinthemodelsituationwillbeapplicabletomore
interestingeverydayproblems.Butofcoursethemainreasonthatdecisiontheoristsstudy
simplegamblesisthatthisiswhatotherdecisiontheoristsdo.
The field had a theory, expected utility theory, which was the foundation of the
rational-agentmodelandistothisdaythemostimportanttheoryinthesocialsciences.
Expected utility theory was not intended as a psychological model; it was a logic of
choice,basedonelementaryrules(axioms)ofrationality.Considerthisexample:
Ifyoupreferanappletoabanana,
then
youalsoprefera10%chancetowinanappletoa10%chancetowinabanana.
The apple and the banana stand for any objects of choice (including gambles), and the
10%chancestandsforanyprobability.ThemathematicianJohnvonNeumann,oneofthe
giantintellectualfiguresofthetwentiethcentury,andtheeconomistOskarMorgenstern
had derived their theory of rational choice between gambles from a few axioms.
Economistsadoptedexpectedutilitytheoryinadualrole:asalogicthatprescribeshow
decisionsshouldbemade,andasadescriptionofhowEconsmakechoices.AmosandI
werepsychologists, however, andwe setout tounderstand how Humansactually make
riskychoices,withoutassuminganythingabouttheirrationality.
We maintained our routine of spending many hours each day in conversation,
sometimesinouroffices,sometimesatrestaurants,oftenonlongwalksthroughthequiet
streetsofbeautifulJerusalem.Aswehaddonewhenwestudiedjudgment,weengagedin
a careful examination of our own intuitive preferences. We spent our time inventing
simpledecisionproblemsandaskingourselveshowwewouldchoose.Forexample:
Whichdoyouprefer?
A.Tossacoin.Ifitcomesupheadsyouwin$100,andifitcomesuptailsyouwin
nothing.
B.Get$46forsure.
WewerenottryingtofigureoutthemosBineithWetrationaloradvantageouschoice;we
wanted to find the intuitive choice, the one that appeared immediately tempting. We
almostalwaysselectedthesameoption.Inthisexample,bothofuswouldhavepickedthe
surething,andyouprobablywoulddothesame.Whenweconfidentlyagreedonachoice,
webelieved—almostalwayscorrectly,asitturnedout—thatmostpeoplewouldshareour
preference,andwemovedonasifwehadsolidevidence.Weknew,ofcourse,thatwe
wouldneedtoverifyourhuncheslater,butbyplayingtherolesofbothexperimentersand
subjectswewereabletomovequickly.
Fiveyearsafterwebeganourstudyofgambles,wefinallycompletedanessaythat
wetitled“ProspectTheory:AnAnalysisofDecisionunderRisk.”Ourtheorywasclosely
modeledonutilitytheorybutdepartedfromitinfundamentalways.Mostimportant,our
model was purely descriptive, and its goal was to document and explain systematic
violations of the axioms of rationality in choices between gambles. We submitted our
essay to Econometrica, a journal that publishes significant theoretical articles in
economicsandindecisiontheory.Thechoiceofvenueturnedouttobeimportant;ifwe
hadpublishedtheidenticalpaperinapsychologicaljournal,itwouldlikelyhavehadlittle
impact on economics. However, our decision was not guided by a wish to influence
economics;Econometricajusthappenedtobewherethebestpapersondecisionmaking
hadbeenpublishedinthepast,andwewereaspiringtobeinthatcompany.Inthischoice
asinmanyothers,wewerelucky.Prospecttheoryturnedouttobethemostsignificant
workweeverdid,andourarticleisamongthemostoftencitedinthesocialsciences.Two
yearslater,wepublishedinScienceanaccountofframingeffects:thelargechangesof
preferencesthataresometimescausedbyinconsequentialvariationsinthewordingofa
choiceproblem.
During the first five years we spent looking at how people make decisions, we
establishedadozenfactsaboutchoicesbetweenriskyoptions.Severalofthesefactswere
inflatcontradictiontoexpectedutilitytheory.Somehadbeenobservedbefore,afewwere
new. Thenwe constructeda theorythat modifiedexpected utilitytheory justenough to
explainourcollectionofobservations.Thatwasprospecttheory.
Our approach to the problem was in the spirit of a field of psychology called
psychophysics, which was founded and named by the German psychologist and mystic
GustavFechner(1801–1887).Fechnerwasobsessedwiththerelationofmindandmatter.
Ononesidethereisaphysicalquantitythatcanvary,suchastheenergyofalight,the
frequency of a tone, or an amount of money. On the other side there is a subjective
experienceofbrightness,pitch,orvalue.Mysteriously,variationsofthephysicalquantity
causevariationsintheintensityorqualityofthesubjectiveexperience.Fechner’sproject
wastofindthepsychophysicallawsthatrelatethesubjectivequantityintheobserver’s
mind to the objective quantity in the material world. He proposed that for many
dimensions,thefunctionislogarithmic—whichsimplymeansthatanincreaseofstimulus
intensitybyagivenfactor(say,times1.5ortimes10)alwaysyieldsthesameincrement
on the psychological scale. If raising the energy of the sound from 10 to 100 units of
physical energy increases psychological intensity by 4 units, then a further increase of
stimulusintensityfrom100to1,000willalsoincreasepsychologicalintensityby4units.
Bernoulli’sError
As Fechner well knew, he was not the first to look for a function that rel
Binepitze=“4”>utility)andtheactualamountofmoney.Hearguedthatagiftof10ducats
has the same utility to someone who already has 100 ducats as a gift of 20 ducats to
someonewhosecurrentwealthis200ducats.Bernoulliwasright,ofcourse:wenormally
speak of changes of income in terms of percentages, as when we say “she got a 30%
raise.”Theideaisthata30%raisemayevokeafairlysimilarpsychologicalresponsefor
therichandforthepoor,whichanincreaseof$100willnotdo.AsinFechner’slaw,the
psychological response to a change of wealth is inversely proportional to the initial
amountofwealth,leadingtotheconclusionthatutilityisalogarithmicfunctionofwealth.
Ifthisfunctionisaccurate,thesamepsychologicaldistanceseparates$100,000from$1
million,and$10millionfrom$100million.
Bernoulli drew on his psychological insight into the utility of wealth to propose a
radically new approach to the evaluation of gambles, an important topic for the
mathematiciansofhisday.PriortoBernoulli,mathematicianshadassumedthatgambles
areassessedbytheirexpectedvalue:aweightedaverageofthepossibleoutcomes,where
eachoutcomeisweightedbyitsprobability.Forexample,theexpectedvalueof:
80%chancetowin$100and20%chancetowin$10is$82(0.8×100+0.2×10).
Nowaskyourselfthisquestion:Whichwouldyouprefertoreceiveasagift,thisgamble
or $80 for sure? Almost everyone prefers the sure thing. If people valued uncertain
prospectsbytheirexpectedvalue,theywouldpreferthegamble,because$82ismorethan
$80.Bernoullipointedoutthatpeopledonotinfactevaluategamblesinthisway.
Bernoulliobservedthatmostpeopledislikerisk(thechanceofreceivingthelowest
possibleoutcome),andiftheyareofferedachoicebetweenagambleandanamountequal
toitsexpectedvaluetheywillpickthesurething.Infactarisk-aversedecisionmakerwill
chooseasurethingthatislessthanexpectedvalue,ineffectpayingapremiumtoavoid
theuncertainty.OnehundredyearsbeforeFechner,Bernoulliinventedpsychophysicsto
explainthisaversiontorisk.Hisideawasstraightforward:people’schoicesarebasednot
on dollar values but on the psychological values of outcomes, their utilities. The
psychological value of a gamble is therefore not the weighted average of its possible
dollaroutcomes;itistheaverageoftheutilitiesoftheseoutcomes,eachweightedbyits
probability.
Table3showsaversionoftheutilityfunctionthatBernoullicalculated;itpresentsthe
utilityofdifferentlevelsofwealth,from1millionto10million.Youcanseethatadding1
million to a wealth of 1 million yields an increment of 20 utility points, but adding 1
million to a wealth of 9 million adds only 4 points. Bernoulli proposed that the
diminishingmarginalvalueofwealth(inthemodernjargon)iswhatexplainsriskaversion
—the common preference that people generally show for a sure thing over a favorable
gambleofequalorslightlyhigherexpectedvalue.Considerthischoice:
Table3
Theexpectedvalueofthegambleandthe“surething”areequalinducats(4million),but
the psychological utilities of the two options are different, because of the diminishing
utilityofwealth:theincrementofutilityfrom1millionto4millionis50units,butan
equalincrement,from4to7million,increasestheutilityofwealthbyonly24units.The
utilityof thegambleis94/2=47(theutilityofitstwooutcomes,each weightedbyits
probability of 1/2). The utility of 4 million is 60. Because 60 is more than 47, an
individualwiththisutilityfunctionwillpreferthesurething.Bernoulli’sinsightwasthata
decisionmakerwithdiminishingmarginalutilityforwealthwillberiskaverse.
Bernoulli’s essay is a marvel of concise brilliance. He applied his new concept of
expectedutility(whichhecalled“moralexpectation”)tocomputehowmuchamerchant
inSt.PetersburgwouldbewillingtopaytoinsureashipmentofspicefromAmsterdamif
“heiswellawareofthefactthatatthistimeofyearofonehundredshipswhichsailfrom
AmsterdamtoPetersburg,fiveareusuallylost.”Hisutilityfunctionexplainedwhypoor
peoplebuyinsuranceandwhyricherpeoplesellittothem.Asyoucanseeinthetable,the
lossof1millioncausesalossof4pointsofutility(from100to96)tosomeonewhohas
10millionandamuchlargerlossof18points(from48to30)tosomeonewhostartsoff
with 3 million. The poorer man will happily pay a premium to transfer the risk to the
richer one, which is what insurance is about. Bernoulli also offered a solution to the
famous “St. Petersburg paradox,” in which people who are offered a gamble that has
infinite expected value (in ducats) are willing to spend only a few ducats for it. Most
impressive,hisanalysisofriskattitudesintermsofpreferencesforwealthhasstoodthe
testoftime:itisstillcurrentineconomicanalysisalmost300yearslater.
Thelongevityofthetheoryisallthemoreremarkablebecauseitisseriouslyflawed.
Theerrorsofatheoryarerarelyfoundinwhatitassertsexplicitly;theyhideinwhatit
ignoresortacitlyassumes.Foranexample,takethefollowingscenarios:
TodayJackandJilleachhaveawealthof5million.
Yesterday,Jackhad1millionandJillhad9million.
Aretheyequallyhappy?(Dotheyhavethesameutility?)
Bernoulli’stheoryassumesthattheutilityoftheirwealthiswhatmakespeoplemoreor
lesshappy.JackandJillhavethesamewealth,andthetheorythereforeassertsthatthey
shouldbeequallyhappy,butyoudonotneedadegreeinpsychologytoknowthattoday
Jack is elated and Jill despondent. Indeed, we know that Jack would be a great deal
happierthanJillevenifhehadonly2milliontodaywhileshehas5.SoBernoulli’stheory
mustbewrong.
The happiness that Jack and Jill experience is determined by the recent change in
theirwealth,relativetothedifferentstatesofwealththatdefinetheirreferencepoints(1
millionforJack,9millionforJill).Thisreferencedependenceisubiquitousinsensation
andperception.Thesamesoundwillbeexperiencedasveryloudorquitefaint,depending
onwhetheritwasprecededbyawhisperorbyaroar.Topredictthesubjectiveexperience
ofloudness,itisnotenoughtoknowitsabsoluteenergy;youalsoneedtoBineli&rquite
faknowthereferencesoundtowhichitisautomaticallycompared.Similarly,youneedto
knowaboutthebackgroundbeforeyoucanpredictwhetheragraypatchonapagewill
appeardarkorlight.Andyouneedtoknowthereferencebeforeyoucanpredicttheutility
ofanamountofwealth.
ForanotherexampleofwhatBernoulli’stheorymisses,considerAnthonyandBetty:
Anthony’scurrentwealthis1million.
Betty’scurrentwealthis4million.
Theyarebothofferedachoicebetweenagambleandasurething.
Thegamble:equalchancestoendupowning1millionor4million
OR
Thesurething:own2millionforsure
In Bernoulli’s account, Anthony and Betty face the same choice: their expected wealth
will be 2.5 million if they take the gamble and 2 million if they prefer the sure-thing
option.BernoulliwouldthereforeexpectAnthonyandBettytomakethesamechoice,but
thispredictionisincorrect.Hereagain,thetheoryfailsbecauseitdoesnotallowforthe
differentreferencepoints from which Anthony and Betty consider their options. If you
imagineyourselfinAnthony’sandBetty’sshoes,youwillquicklyseethatcurrentwealth
mattersagreatdeal.Hereishowtheymaythink:
Anthony(whocurrentlyowns1million):“IfIchoosethesurething,mywealthwill
doublewithcertainty.Thisisveryattractive.Alternatively,Icantakeagamblewith
equalchancestoquadruplemywealthortogainnothing.”
Betty(whocurrentlyowns4million):“IfIchoosethesurething,Ilosehalfofmy
wealthwithcertainty,whichisawful.Alternatively,Icantakeagamblewithequal
chancestolosethree-quartersofmywealthortolosenothing.”
YoucansensethatAnthonyandBettyarelikelytomakedifferentchoicesbecausethe
sure-thingoptionofowning2millionmakesAnthonyhappyandmakesBettymiserable.
Note also how the sure outcome differs from the worst outcome of the gamble: for
Anthony,itisthedifferencebetweendoublinghiswealthandgainingnothing;forBetty,it
isthe difference betweenlosing halfher wealthand losing three-quartersof it.Bettyis
muchmorelikelytotakeherchances,asothersdowhenfacedwithverybadoptions.AsI
have told their story, neither Anthony nor Betty thinks in terms of states of wealth:
Anthony thinks of gains and Betty thinks of losses. The psychological outcomes they
assessareentirelydifferent,althoughthepossiblestatesofwealththeyfacearethesame.
BecauseBernoulli’smodellackstheideaofareferencepoint,expectedutilitytheory
doesnotrepresenttheobviousfactthattheoutcomethatisgoodforAnthonyisbadfor
Betty.HismodelcouldexplainAnthony’sriskaversion,butitcannotexplainBetty’srisk-
seekingpreferenceforthegamble,abehaviorthatisoftenobservedinentrepreneursand
ingeneralswhenalltheiroptionsarebad.
All this is rather obvious, isn’t it? One could easily imagine Bernoulli himself
constructing similar examples and developing a more complex theory to accommodate
them; for some reason, he did not. One could also imagine colleagues of his time
disagreeingwithhim,orlaterscholarsobjectingastheyreadhisessay;forsomereason,
theydidnoteither.
Themysteryishowaconceptionoftheutilityofoutcomesthatisvulnerabletosuch
obviouscounterexamplessurvivedforsolong.Icanexplainitonlybyaweaknessofthe
scholarly mind that I have often observed in myself. I call it theory-induced blindness:
onceyouhaveacceptedatheoryanduseditasatoolinyourthinking,itisextraordinarily
difficulttonoticeitsflaws.Ifyoucomeuponanobservationthatdoesnotseemtofitthe
model,youassumethattheremustbeaperfectlygoodexplanationthatyouaresomehow
missing.Yougivethetheorythebenefitofthedoubt,trustingthecommunityofexperts
whohaveacceptedit.Manyscholarshavesurelythoughtatonetimeoranotherofstories
suchasthoseofAnthonyandBetty,orJackandJill,andcasuallynotedthatthesestories
didnot jibe withutilitytheory.But theydidnot pursuethe idea tothe point ofsaying,
“This theory is seriously wrong because it ignores the fact that utility depends on the
historyofone’swealth,notonlyonpresentwealth.”AsthepsychologistDanielGilbert
observed,disbelievingishardwork,andSystem2iseasilytired.
SpeakingofBernoulli’sErrors
“Hewasveryhappywitha$20,000bonusthreeyearsago,buthissalaryhasgoneup
by20%since,sohewillneedahigherbonustogetthesameutility.”
“Bothcandidatesare willingtoacceptthesalarywe’reoffering,buttheywon’tbe
equally satisfied because their reference points are different. She currently has a
muchhighersalary.”
“She’ssuinghimforalimony.Shewouldactuallyliketosettle,buthepreferstogoto
court.That’snotsurprising—shecanonlygain,soshe’sriskaverse.He,ontheother
hand,facesoptionsthatareallbad,sohe’drathertaketherisk.”
P
ProspectTheory
AmosandIstumbledonthecentralflawinBernoulli’stheorybyaluckycombinationof
skill and ignorance. At Amos’s suggestion, I read a chapter in his book that described
experimentsinwhichdistinguishedscholarshadmeasuredtheutilityofmoneybyasking
peopletomakechoicesaboutgamblesinwhichtheparticipantcouldwinorloseafew
pennies. The experimenters were measuring the utility of wealth, by modifying wealth
withinarangeoflessthanadollar.Thisraisedquestions.Isitplausibletoassumethat
peopleevaluatethegamblesbytinydifferencesinwealth?Howcouldonehopetolearn
aboutthepsychophysicsofwealthbystudyingreactionstogainsandlossesofpennies?
Recent developments in psychophysical theory suggested that if you want to study the
subjectivevalueofwealth,youshouClth”ldaskdirectquestionsaboutwealth,notabout
changesofwealth.Ididnotknowenoughaboututilitytheorytobeblindedbyrespectfor
it,andIwaspuzzled.
WhenAmosandImetthenextday,Ireportedmydifficultiesasavaguethought,not
asadiscovery.Ifullyexpectedhimtosetmestraightandtoexplainwhytheexperiment
thathadpuzzledmemadesenseafterall,buthedidnothingofthekind—therelevanceof
the modern psychophysics was immediately obvious to him. He remembered that the
economist Harry Markowitz, who would later earn the Nobel Prize for his work on
finance,hadproposedatheoryinwhichutilitieswereattachedtochangesofwealthrather
thantostatesofwealth.Markowitz’sideahadbeenaroundforaquarterofacenturyand
hadnotattractedmuchattention,butwequicklyconcludedthatthiswasthewaytogo,
and that the theory we were planning to develop would define outcomes as gains and
losses, not as states of wealth. Knowledge of perception and ignorance about decision
theorybothcontributedtoalargestepforwardinourresearch.
We soon knew that we had overcome a serious case of theory-induced blindness,
becausetheideawehadrejectednowseemednotonlyfalsebutabsurd.Wewereamused
to realize that we were unable to assess our current wealth withintens of thousands of
dollars. The idea of deriving attitudes to small changes from the utility of wealth now
seemedindefensible.Youknowyouhave made atheoreticaladvancewhenyoucan no
longerreconstructwhyyoufailedforsolongtoseetheobvious.Still,ittookusyearsto
exploretheimplicationsofthinkingaboutoutcomesasgainsandlosses.
In utility theory, the utility of a gain is assessed by comparing the utilities of two
statesofwealth.Forexample,theutilityofgettinganextra$500whenyourwealthis$1
millionis the differencebetween theutility of $1,000,500 and theutility of$1 million.
And if you own the larger amount, the disutility of losing $500 is again the difference
betweentheutilitiesofthetwostatesofwealth.Inthistheory,theutilitiesofgainsand
lossesareallowedtodifferonlyintheirsign(+or–).Thereisnowaytorepresentthefact
that the disutility of losing $500 could be greater than the utility of winning the same
amount—though of course it is. As might be expected in a situation of theory-induced
blindness, possible differences between gains and losses were neither expected nor
studied.Thedistinctionbetweengainsandlosseswasassumednottomatter,sotherewas
nopointinexaminingit.
AmosandIdidnotseeimmediatelythatourfocusonchangesofwealthopenedthe
waytoanexplorationofanewtopic.Weweremainlyconcernedwithdifferencesbetween
gambles with high or low probability of winning. One day, Amos made the casual
suggestion,“Howaboutlosses?”andwequicklyfoundthatourfamiliarriskaversionwas
replacedbyriskseekingwhenweswitchedourfocus.Considerthesetwoproblems:
Problem1:Whichdoyouchoose?
Get$900forsureOR90%chancetoget$1,000
Problem2:Whichdoyouchoose?
Lose$900forsureOR90%chancetolose$1,000
You were probably risk averse in problem 1, as is the great majority of people. The
subjectivevalueofagainof$900iscertainlymorethan90%ofthevalueofagaBlth”it
ueofagin of$1,000.Therisk-aversechoice in thisproblemwouldnot have surprised
Bernoulli.
Nowexamineyourpreferenceinproblem2.Ifyouarelikemostotherpeople,you
chose the gamble in this question. The explanation for this risk-seeking choice is the
mirror image of the explanation of risk aversion in problem 1: the (negative) value of
losing$900ismuchmorethan90%ofthe(negative)valueoflosing$1,000.Thesureloss
isveryaversive,andthisdrivesyoutotaketherisk.Later,wewillseethattheevaluations
oftheprobabilities(90%versus100%)alsocontributestobothriskaversioninproblem1
andthepreferenceforthegambleinproblem2.
Wewerenotthefirsttonoticethatpeoplebecomeriskseekingwhenalltheiroptions
arebad,buttheory-inducedblindnesshadprevailed.Becausethedominanttheorydidnot
provideaplausiblewaytoaccommodatedifferentattitudestoriskforgainsandlosses,the
fact that the attitudes differed had to be ignored. In contrast, our decision to view
outcomes as gains and losses led us to focus precisely on this discrepancy. The
observationofcontrastingattitudestoriskwithfavorableandunfavorableprospectssoon
yielded a significant advance: we found a way to demonstrate the central error in
Bernoulli’smodelofchoice.Havealook:
Problem3:Inadditiontowhateveryouown,youhavebeengiven$1,000.
Youarenowaskedtochooseoneoftheseoptions:
50%chancetowin$1,000ORget$500forsure
Problem4:Inadditiontowhateveryouown,youhavebeengiven$2,000.
Youarenowaskedtochooseoneoftheseoptions:
50%chancetolose$1,000ORlose$500forsure
You can easily confirm that in terms of final states of wealth—all that matters for
Bernoulli’s theory—problems 3 and 4 are identical. In both cases you have a choice
between the same two options: you can have the certainty of being richer than you
currentlyareby$1,500,oracceptagambleinwhichyouhaveequalchancestobericher
by$1,000orby$2,000.InBernoulli’stheory,therefore,thetwoproblemsshouldelicit
similarpreferences.Checkyourintuitions,andyouwillprobablyguesswhatotherpeople
did.
Inthefirstchoice,alargemajorityofrespondentspreferredthesurething.
Inthesecondchoice,alargemajoritypreferredthegamble.
The finding of different preferences in problems 3 and 4 was a decisive
counterexample to the key idea of Bernoulli’s theory. If the utility of wealth is all that
matters, then transparently equivalent statements of the same problem should yield
identicalchoices.Thecomparisonoftheproblemshighlightstheall-importantroleofthe
referencepointfromwhichtheoptionsareevaluated.Thereferencepointishigherthan
currentwealthby$1,000inproblem3,by$2,000inproblem4.Beingricherby$1,500is
thereforeagainof$500inproblem3andalossinproblem4.Obviously,otherexamples
of the same kind are easy to generate. The story of Anthony and Betty had a similar
structure.
Howmuchattentiondidyoupaytothegiftof$1,000or$2,000thatyouwere“given”
priortomakingyourchoice?Ifyouarelikemostpeople,youbarelynoticedit.Indeed,
therewasnoreasonforyoutoattendtoit,becausethegiftisincludedinthereference
point, and reference points are generally ignored. You know something about your
preferencesthatutilitytheoristsdonot—thatyourattitudestoriskwouldnotbedifferent
ifyournetworthwerehigherorlowerbyafewthousanddollars(unlessyouareabjectly
poor).Andyoualsoknowthatyourattitudestogainsandlossesarenotderivedfromyour
evaluationofyourwealth.Thereasonyouliketheideaofgaining$100anddislikethe
ideaoflosing$100isnotthattheseamountschangeyourwealth.Youjustlikewinning
anddislikelosing—andyoualmostcertainlydislikelosingmorethanyoulikewinning.
The four problems highlight the weakness of Bernoulli’s model. His theory is too
simpleand lacksa movingpart. The missing variable is the referencepoint, the earlier
staterelativetowhichgainsandlossesareevaluated.InBernoulli’stheoryyouneedto
knowonlythestateofwealthtodetermineitsutility,butinprospecttheoryyoualsoneed
toknowthereferencestate.Prospecttheoryisthereforemorecomplexthanutilitytheory.
Insciencecomplexityisconsideredacost,whichmustbejustifiedbyasufficientlyrich
setofnewand(preferably)interestingpredictionsoffactsthattheexistingtheorycannot
explain.Thiswasthechallengewehadtomeet.
AlthoughAmosandIwerenotworkingwiththetwo-systemsmodelofthemind,it’s
clearnowthattherearethreecognitivefeaturesattheheartofprospecttheory.Theyplay
an essential role in the evaluation of financial outcomes and are common to many
automatic processes of perception, judgment, and emotion. They should be seen as
operatingcharacteristicsofSystem1.
Evaluationisrelativetoaneutralreferencepoint,whichissometimesreferredtoas
an “adaptation level.” You can easily set up a compelling demonstration of this
principle.Placethreebowlsofwaterinfrontofyou.Puticewaterintotheleft-hand
bowlandwarmwaterintotheright-handbowl.Thewaterinthemiddlebowlshould
beatroomtemperature.Immerseyourhandsinthecoldandwarmwaterforabouta
minute,thendipbothinthemiddlebowl.Youwillexperiencethesametemperature
asheatinonehandandcoldintheother.Forfinancialoutcomes,theusualreference
pointisthestatusquo,butitcanalsobetheoutcomethatyouexpect,orperhapsthe
outcome to which you feel entitled, for example, the raise or bonus that your
colleagues receive. Outcomes that are better than the reference points are gains.
Belowthereferencepointtheyarelosses.
A principle of diminishing sensitivity applies to both sensory dimensions and the
evaluationofchangesofwealth.Turningonaweaklighthasalargeeffectinadark
room. The same increment of light may be undetectable in a brightly illuminated
room.Similarly,thesubjectivedifferencebetween$900and$1,000ismuchsmaller
thanthedifferencebetween$100and$200.
The third principle is loss aversion. When directly compared or weighted against
each other, losses loom larger than gains. This asymmetry between the power of
positive and negative expectations or experiences has an evolutionary history.
Organismsthattreatthreatsasmoreurgentthanopportunitieshaveabetterchanceto
surviveandreproduce.
The three principles that govern the value of outcomes are illustrated by figure 1
Blth”wagure0.Ifprospecttheoryhadaflag,thisimagewouldbedrawnonit.Thegraph
showsthe psychological value of gainsand losses, which are the“carriers” of value in
prospect theory (unlike Bernoulli’s model, in which states of wealth are the carriers of
value).Thegraphhastwodistinctparts,totherightandtotheleftofaneutralreference
point.AsalientfeatureisthatitisS-shaped,whichrepresentsdiminishingsensitivityfor
bothgainsandlosses.Finally,thetwocurvesoftheSarenotsymmetrical.Theslopeof
thefunctionchangesabruptlyatthereferencepoint:theresponsetolossesisstrongerthan
theresponsetocorrespondinggains.Thisislossaversion.
Figure10
LossAversion
Manyoftheoptionswefaceinlifeare“mixed”:thereisariskoflossandanopportunity
for gain, and we must decide whether to accept the gamble or reject it. Investors who
evaluateastart-up,lawyerswhowonderwhethertofilealawsuit,wartimegeneralswho
consideranoffensive,andpoliticianswhomustdecidewhethertorunforofficeallface
the possibilities of victory or defeat. For an elementary example of a mixed prospect,
examineyourreactiontothenextquestion.
Problem5:Youareofferedagambleonthetossofacoin.
Ifthecoinshowstails,youlose$100.
Ifthecoinshowsheads,youwin$150.
Isthisgambleattractive?Wouldyouacceptit?
Tomakethischoice,youmustbalancethepsychologicalbenefitofgetting$150against
thepsychologicalcostoflosing$100.Howdoyoufeelaboutit?Althoughtheexpected
valueofthegambleisobviouslypositive,becauseyoustandtogainmorethanyoucan
lose,you probablydislikeit—mostpeopledo.Therejectionofthisgambleisanactof
System2,butthecriticalinputsareemotionalresponsesthataregeneratedbySystem1.
Formostpeople,thefearoflosing$100ismoreintensethanthehopeofgaining$150.
Weconcludedfrommanysuchobservationsthat“lossesloomlargerthangains”andthat
peoplearelossaverse.
Youcanmeasuretheextentofyouraversiontolossesbyaskingyourselfaquestion:
WhatisthesmallestgainthatIneedtobalanceanequalchancetolose$100?Formany
peopletheanswerisabout$200,twiceasmuchastheloss.The“lossaversionratio”has
beenestimatedinseveralexperimentsandisusuallyintherangeof1.5to2.5.Thisisan
average,ofcourse;somepeoplearemuchmorelossaversethanothers.Professionalrisk
takersinthefinancialmarketsaremoretolerantoflosses,probablybecausetheydonot
respond emotionally to every fluctuation. When participants in an experiment were
instructed to “think like a trader,” they became less loss averse and their emotional
reactiontolosses(measuredbyaphysiologicalindexofemotionalarousal)wassharply
reduced.
In order to examine your loss aversion ratio for different stakes, consider the
following questions. Ignore any social considerations, do not try to appear either bold
Blth”vioherorcautious,andfocusonlyonthesubjectiveimpactofthepossiblelossand
theoffsettinggain.
Considera50–50gambleinwhichyoucanlose$10.Whatisthesmallestgainthat
makesthegambleattractive?Ifyousay$10,thenyouareindifferenttorisk.Ifyou
giveanumberlessthan$10,youseekrisk.Ifyouranswerisabove$10,youareloss
averse.
Whataboutapossiblelossof$500onacointoss?Whatpossiblegaindoyourequire
tooffsetit?
Whataboutalossof$2,000?
Asyoucarriedoutthis exercise, youprobablyfoundthatyour loss aversioncoefficient
tendstoincreasewhenthestakesrise,butnotdramatically.Allbetsareoff,ofcourse,if
thepossiblelossispotentiallyruinous,orifyourlifestyleisthreatened.Thelossaversion
coefficientisverylargeinsuchcasesandmayevenbeinfinite—therearerisksthatyou
willnotaccept,regardlessofhowmanymillionsyoumightstandtowinifyouarelucky.
Anotherlook atfigure 10 mayhelp preventa commonconfusion. In thischapter I
havemadetwoclaims,whichsomereadersmayviewascontradictory:
Inmixed gambles, where both again and aloss are possible, loss aversioncauses
extremelyrisk-aversechoices.
Inbadchoices,whereasurelossiscomparedtoalargerlossthatismerelyprobable,
diminishingsensitivitycausesriskseeking.
Thereisnocontradiction.Inthemixedcase,thepossiblelossloomstwiceaslargeasthe
possiblegain,asyoucanseebycomparingtheslopesofthevaluefunctionforlossesand
gains.Inthebadcase,thebendingofthevaluecurve(diminishingsensitivity)causesrisk
seeking.Thepainoflosing$900ismorethan90%ofthepainoflosing$1,000.Thesetwo
insightsaretheessenceofprospecttheory.
Figure10showsanabruptchangeintheslopeofthevaluefunctionwheregainsturninto
losses, because there is considerable loss aversion even when the amount at risk is
minusculerelativetoyour wealth.Isitplausible that attitudestostates of wealthcould
explain the extreme aversion to small risks? It is a striking example of theory-induced
blindnessthatthisobviousflawinBernoulli’stheoryfailedtoattractscholarlynoticefor
more than 250 years. In 2000, the behavioral economist Matthew Rabin finally proved
mathematicallythatattemptstoexplainlossaversionbytheutilityofwealthareabsurd
anddoomedtofail,andhisproofattractedattention.Rabin’stheoremshowsthatanyone
whorejectsafavorablegamblewithsmallstakesismathematicallycommittedtoafoolish
levelofriskaversionforsomelargergamble.Forexample,henotesthatmostHumans
rejectthefollowinggamble:
50%chancetolose$100and50%chancetowin$200
Hethenshowsthataccordingtoutilitytheory,anindividualwhorejectsthatgamblewill
alsoturndownthefollowinggamble:
50%chancetolose$200and50%chancetowin$20,000
But of course no one in his or her right mind will reject this gamble! In an exuberant
articletheywroteaboBlth”ins>
Perhapscarriedawaybytheirenthusiasm,theyconcludedtheirarticlebyrecallingthe
famous Monty Python sketch in which a frustrated customer attempts to return a dead
parrottoapetstore.Thecustomerusesalongseriesofphrasestodescribethestateofthe
bird,culminatingin“thisisanex-parrot.”RabinandThalerwentontosaythat“itistime
foreconomiststorecognizethatexpectedutilityisanex-hypothesis.”Manyeconomists
saw this flippant statement as little short of blasphemy. However, the theory-induced
blindnessofacceptingtheutilityofwealthasanexplanationofattitudestosmalllossesis
alegitimatetargetforhumorouscomment.
BlindSpotspfProspectTheory
SofarinthispartofthebookIhaveextolledthevirtuesofprospecttheoryandcriticized
therationalmodelandexpectedutilitytheory.Itistimeforsomebalance.
Most graduate students in economics have heard about prospect theory and loss
aversion,butyouareunlikelytofindthesetermsintheindexofanintroductorytextin
economics. I am sometimes pained by this omission, but in fact it is quite reasonable,
becauseofthecentralroleofrationalityinbasiceconomictheory.Thestandardconcepts
and results that undergraduates are taught are most easily explained by assuming that
Econsdonotmakefoolishmistakes.Thisassumptionistrulynecessary,anditwouldbe
undermined by introducing the Humans of prospect theory, whose evaluations of
outcomesareunreasonablyshort-sighted.
There are good reasons for keeping prospect theory out of introductory texts. The
basicconceptsofeconomicsareessentialintellectualtools,whicharenoteasytograsp
evenwithsimplifiedandunrealisticassumptionsaboutthenatureoftheeconomicagents
who interact in markets. Raising questions about these assumptions even as they are
introducedwouldbeconfusing,andperhapsdemoralizing.Itisreasonabletoputpriority
onhelpingstudentsacquirethe basic tools ofthediscipline.Furthermore,thefailureof
rationality that is built into prospect theory is often irrelevant to the predictions of
economictheory,whichworkoutwithgreatprecisioninsomesituationsandprovidegood
approximations in many others. In some contexts, however, the difference becomes
significant: the Humans described by prospect theory are guided by the immediate
emotional impact of gains and losses, not by long-term prospects of wealth and global
utility.
I emphasized theory-induced blindness in my discussion of flaws in Bernoulli’s
model that remained unquestioned for more than two centuries. But of course theory-
inducedblindnessisnotrestrictedtoexpectedutilitytheory.Prospecttheoryhasflawsof
itsown,andtheory-inducedblindnesstotheseflawshascontributedtoitsacceptanceas
themainalternativetoutilitytheory.
Considertheassumptionofprospecttheory,thatthereferencepoint,usuallythestatus
quo,hasavalueofzero.Thisassumptionseemsreasonable,butitleadstosomeabsurd
consequences.Haveagoodlookatthefollowingprospects.Whatwoulditbeliketoown
them?
A.onechanceinamilliontowin$1million
B.90%chancetowin$12and10%chancetowinnothing
C.90%chancetowin$1millionand10%chancetowinnothing
Winningnothingisapossibleoutcomeinallthreegambles,andprospecttheoryassigns
thesamevaluetothatoutcomeinthethreecases.Winningnothingisthereferencepoint
anditsvalueiszero.Dothesestatementscorrespondtoyourexperience?Ofcoursenot.
Winning nothing is a nonevent in the first two cases, and assigning it a value of zero
makes good sense. In contrast, failing to win in the third scenario is intensely
disappointing. Like a salary increase that has been promised informally, the high
probabilityofwinningthelargesumsetsupatentativenewreferencepoint.Relativeto
your expectations, winning nothing will be experienced as a large loss. Prospect theory
cannotcopewiththisfact,becauseitdoesnotallowthevalueofanoutcome(inthiscase,
winning nothing) to change when it is highly unlikely, or when the alternative is very
valuable. In simple words, prospect theory cannot deal with disappointment.
Disappointmentandtheanticipationofdisappointmentarereal,however,andthefailure
to acknowledge them is as obvious a flow as the counterexamples that I invoked to
criticizeBernoulli’stheory.
Prospecttheoryandutilitytheoryalsofailtoallowforregret.Thetwotheoriesshare
the assumption that available options in a choice are evaluated separately and
independently,andthattheoptionwiththehighestvalueisselected.Thisassumptionis
certainlywrong,asthefollowingexampleshows.
Problem6:Choosebetween90%chancetowin$1millionOR$50withcertainty.
Problem 7: Choose between 90% chance to win $1 million OR $150,000 with
certainty.
Comparetheanticipatedpainofchoosingthegambleandnotwinninginthetwocases.
Failing to win is a disappointment in both, but the potential pain is compounded in
problem7byknowingthatifyouchoosethegambleandloseyouwillregretthe“greedy”
decision you made by spurning a sure gift of $150,000. In regret, the experience of an
outcomedependsonanoptionyoucouldhaveadoptedbutdidnot.
Severaleconomistsandpsychologistshaveproposedmodelsofdecisionmakingthat
arebasedontheemotionsofregretanddisappointment.Itisfairtosaythatthesemodels
havehadlessinfluencethanprospecttheory,andthereasonisinstructive.Theemotions
of regret and disappointment are real, and decision makers surely anticipate these
emotions when making their choices. The problem is that regret theories make few
striking predictions that would distinguish them from prospect theory, which has the
advantageofbeingsimpler.Thecomplexityofprospecttheorywasmoreacceptableinthe
competitionwithexpectedutilitytheorybecauseitdidpredictobservationsthatexpected
utilitytheorycouldnotexplain.
Richer and more realistic assumptions do not suffice to make a theory successful.
Scientistsusetheoriesasabagofworkingtools,andtheywillnottakeontheburdenofa
heavierbagunlessthenewtoolsareveryuseful.Prospecttheorywasacceptedbymany
scholarsnotbecauseitis“true”butbecausetheconceptsthatitaddedtoutilitytheory,
notablythereferencepoint and loss aversion,wereworththetrouble;theyyieldednew
predictionsthatturnedouttobetrue.Wewerelucky.
SpeakingofProspectTheory
“He suffers from extreme loss aversion, which makes him turn down very favorable
opportunities.”
“Consideringhervastwealth,heremotionalresponsetotrivialgainsandlossesmakesno
sense.”
“Heweighslossesabouttwiceasmuchasgains,whichisnormal.”
P
TheEndowmentEffect
Youhaveprobablyseenfigure11oraclosecousinofitevenifyouneverhadaclassin
economics.Thegraphdisplaysanindividual’s“indifferencemap”fortwogoods.
Figure11
Studentslearninintroductoryeconomicsclassesthateachpointonthemapspecifies
aparticularcombinationofincomeandvacationdays.Each“indifferencecurve”connects
thecombinationsofthetwogoodsthatareequallydesirable—theyhavethesameutility.
Thecurveswouldturnintoparallelstraightlinesifpeoplewerewillingto“sell”vacation
daysforextraincomeatthesamepriceregardlessofhowmuchincomeandhowmuch
vacation time they have. The convex shape indicates diminishing marginal utility: the
moreleisureyouhave,thelessyoucareforanextradayofit,andeachaddeddayisworth
lessthantheonebefore.Similarly,themoreincomeyou have, thelessyoucareforan
extra dollar, and the amount you are willing to give up for an extra day of leisure
increases.
All locations on an indifference curve are equally attractive. This is literally what
indifferencemeans:youdon’tcarewhereyouareonanindifferencecurve.SoifAandB
areonthesameindifferencecurveforyou,youareindifferentbetweenthemandwillneed
no incentive to move from one to the other, or back. Some version of this figure has
appeared in every economics textbook written in the last hundred years, and many
millionsofstudentshavestaredatit.Fewhavenoticedwhatismissing.Hereagain,the
powerandeleganceofatheoreticalmodelhaveblindedstudentsandscholarstoaserious
deficiency.
Whatis missingfrom thefigure isan indicationof theindividual’s currentincome
andleisure.Ifyouareasalariedemployee,thetermsofyouremploymentspecifyasalary
andanumberofvacationdays,whichisapointonthemap.Thisisyourreferencepoint,
yourstatusquo,butthefiguredoesnotshowit.Byfailingtodisplayit,thetheoristswho
drawthisfigureinviteyoutobelievethatthereferencepointdoesnotmatter,butbynow
youknowthatofcourseitdoes.ThisisBernoulli’serroralloveragain.Therepresentation
of indifference curves implicitly assumes that your utility at any given moment is
determined entirely by your present situation, that the past is irrelevant, and that your
evaluation of a possible job does not depend on the terms of your current job. These
assumptionsarecompletelyunrealisticinthiscaseandinmanyothers.
TheomissionoftherefConserencepointfromtheindifferencemapisasurprising
case of theory-induced blindness, because we so often encounter cases in which the
referencepointobviouslymatters.Inlabornegotiations,itiswellunderstoodbybothsides
that the reference point is the existing contract and that the negotiations will focus on
mutualdemandsforconcessionsrelativetothatreferencepoint.Theroleoflossaversion
inbargainingisalsowellunderstood:makingconcessionshurts.Youhavemuchpersonal
experience of the role of reference point. If you changed jobs or locations, or even
consideredsuchachange,yousurelyrememberthatthefeaturesofthenewplacewere
codedasplusesorminusesrelativetowhereyouwere.Youmayalsohavenoticedthat
disadvantages loomed larger than advantages in this evaluation—loss aversion was at
work.Itisdifficulttoacceptchangesfortheworse.Forexample,theminimalwagethat
unemployedworkerswouldacceptfornewemploymentaverages90%oftheirprevious
wage,anditdropsbylessthan10%overaperiodofoneyear.
To appreciate the power that the reference point exerts on choices, consider Albert
andBen,“hedonictwins”whohaveidenticaltastesandcurrentlyholdidenticalstarting
jobs,withlittleincomeandlittleleisuretime.Theircurrentcircumstancescorrespondto
thepointmarked1infigure11.Thefirmoffersthemtwoimprovedpositions,AandB,
andletsthemdecidewhowillgetaraiseof$10,000(positionA)andwhowillgetanextra
day of paid vacation each month (position B). As they are both indifferent, they toss a
coin.Albertgetstheraise,Bengetstheextraleisure.Sometimepassesasthetwinsget
accustomedtotheir positions.Nowthe companysuggeststhey mayswitchjobs ifthey
wish.
Thestandardtheoryrepresentedinthefigureassumesthatpreferencesarestableover
time.PositionsAandBareequallyattractiveforbothtwinsandtheywillneedlittleorno
incentive to switch. In sharp contrast, prospect theory asserts that both twins will
definitelyprefertoremainastheyare.Thispreferenceforthestatusquoisaconsequence
oflossaversion.
Let us focus on Albert. He was initially in position 1 on the graph, and from that
referencepointhefoundthesetwoalternativesequallyattractive:
GotoA:araiseof$10,000
OR
GotoB:12extradaysofvacation
TakingpositionAchangesAlbert’sreferencepoint,andwhenheconsidersswitchingtoB,
hischoicehasanewstructure:
StayatA:nogainandnoloss
OR
MovetoB:12extradaysofvacationanda$10,000salarycut
Youjusthadthesubjectiveexperienceoflossaversion.Youcouldfeelit:asalarycutof
$10,000isverybadnews.Evenifagainof12vacationdayswasasimpressiveasagain
of$10,000,thesameimprovementofleisureisnotsufficienttocompensateforalossof
$10,000. Albert will stay at A because the disadvantage of moving outweighs the
advantage.ThesamereasoningappliestoBen,whowillalsowanttokeephispresentjob
becausethelossofnow-preciousleisureoutweighsthebenefitoftheextraincome.
This example highlights two aspects of choice that the st Bon s Ae st Bonandard
modelofindifferencecurvesdoesnotpredict.First,tastesarenotfixed;theyvarywiththe
referencepoint.Second,the disadvantages ofachange loom largerthanitsadvantages,
inducingabiasthatfavorsthestatusquo.Ofcourse,lossaversiondoesnotimplythatyou
never prefer to change your situation; the benefits of an opportunity may exceed even
overweightedlosses.Lossaversionimpliesonlythatchoicesarestronglybiasedinfavor
ofthereferencesituation(andgenerallybiasedtofavorsmallratherthanlargechanges).
ConventionalindifferencemapsandBernoulli’srepresentationofoutcomesasstates
ofwealthshareamistakenassumption:thatyourutilityforastateofaffairsdependsonly
onthatstateandisnotaffectedbyyourhistory.Correctingthatmistakehasbeenoneof
theachievementsofbehavioraleconomics.
TheEndowmentEffect
Thequestionofwhenanapproachoramovementgotitsstartisoftendifficulttoanswer,
buttheoriginofwhatisnowknownasbehavioraleconomicscanbespecifiedprecisely.
In the early 1970s, Richard Thaler, then a graduate student in the very conservative
economics department of the University of Rochester, began having heretical thoughts.
Thaleralwayshadasharpwitandanironicbent,andasastudentheamusedhimselfby
collectingobservationsofbehaviorthatthemodelofrationaleconomicbehaviorcouldnot
explain. He took special pleasure in evidence of economic irrationality among his
professors,andhefoundonethatwasparticularlystriking.
ProfessorR(nowrevealedtobeRichardRosett,whowentontobecomethedeanof
theUniversityofChicagoGraduateSchoolofBusiness)wasafirmbelieverinstandard
economictheoryaswellasasophisticatedwinelover.ThalerobservedthatProfessorR
wasveryreluctanttosellabottlefromhiscollection—evenatthehighpriceof$100(in
1975dollars!).ProfessorRboughtwineatauctions,butwouldneverpaymorethan$35
forabottleofthatquality.Atpricesbetween$35and$100,hewouldneitherbuynorsell.
Thelargegapisinconsistentwitheconomictheory,inwhichtheprofessorisexpectedto
haveasinglevalueforthebottle.Ifaparticularbottleisworth$50tohim,thenheshould
bewillingtosellitforanyamountinexcessof$50.Ifhedidnotownthebottle,heshould
bewillingtopayanyamountupto$50forit.Thejust-acceptablesellingpriceandthe
just-acceptablebuyingpriceshouldhavebeenidentical,butinfacttheminimumpriceto
sell($100)wasmuchhigherthanthemaximumbuyingprice of $35.Owningthegood
appearedtoincreaseitsvalue.
Richard Thaler found many examples of what he called the endowment effect,
especially for goods that are not regularly traded. You can easily imagine yourself in a
similarsituation.Supposeyouholdatickettoasold-outconcertbyapopularband,which
youboughtattheregularpriceof$200.Youareanavidfanandwouldhavebeenwilling
topayupto$500fortheticket.NowyouhaveyourticketandyoulearnontheInternet
thatricherormoredesperatefansareoffering$3,000.Wouldyousell?Ifyouresemble
mostoftheaudienceatsold-outeventsyoudonotsell.Yourlowestsellingpriceisabove
$3,000 and your maximum buying price is $500. This is an example of an endowment
effect, and a believer in standard economic theory would be puzzled by it. Thaler was
lookingforanaccountthatcouldexplainpuzzlesofthiskind.
ChanceintervenedwhenThalermetoneofourformerstudentsataconferenceand
obtained an early draft of prospect theory. He reports that he read the manuscript with
considerableBonsAbleBonexcitement,becausehequicklyrealizedthattheloss-averse
value function of prospect theory could explain the endowment effect and some other
puzzlesinhiscollection.ThesolutionwastoabandonthestandardideathatProfessorR
hadauniqueutilityforthestateofhavingaparticularbottle.Prospecttheorysuggested
thatthewillingnesstobuyorsellthebottledependsonthereferencepoint—whetheror
nottheprofessorownsthebottlenow.Ifheownsit,heconsidersthepainofgivingupthe
bottle.Ifhedoesnotownit,heconsidersthepleasureofgettingthebottle.Thevalues
wereunequalbecauseoflossaversion:givingupabottleofnicewineismorepainfulthan
gettinganequallygoodbottleispleasurable.Rememberthegraphoflossesandgainsin
the previous chapter. The slope of the function is steeper in the negative domain; the
response to a loss is stronger than the response to a corresponding gain. This was the
explanation of the endowment effect that Thaler had been searching for. And the first
application of prospect theory to an economic puzzle now appears to have been a
significantmilestoneinthedevelopmentofbehavioraleconomics.
ThalerarrangedtospendayearatStanfordwhenheknewthatAmosandIwouldbe
there. During this productive period, we learned much from each other and became
friends.Sevenyearslater,heandIhadanotheropportunitytospendayeartogetherandto
continue the conversation between psychology and economics. The Russell Sage
Foundation,whichwasforalongtimethemainsponsorofbehavioraleconomics,gave
oneofitsfirstgrantstoThalerforthepurposeofspendingayearwithmeinVancouver.
Duringthatyear,weworkedcloselywithalocaleconomist,JackKnetsch,withwhomwe
sharedintenseinterestintheendowmenteffect,therulesofeconomicfairness,andspicy
Chinesefood.
The starting point for our investigation was that the endowment effect is not
universal.Ifsomeoneasksyoutochangea$5billforfivesingles,youhandoverthefive
oneswithoutanysenseofloss.Noristheremuchlossaversionwhenyoushopforshoes.
The merchant who gives up the shoes in exchange for money certainly feels no loss.
Indeed, the shoes that he hands over have always been, from his point of view, a
cumbersome proxy for money that he was hoping to collect from some consumer.
Furthermore,youprobablydonotexperiencepayingthemerchantasaloss,becauseyou
wereeffectivelyholdingmoneyasaproxyfortheshoesyouintendedtobuy.Thesecases
of routine trading are not essentially different from the exchange of a $5 bill for five
singles.Thereisnolossaversiononeithersideofroutinecommercialexchanges.
WhatdistinguishesthesemarkettransactionsfromProfessorR’sreluctancetosellhis
wine,orthereluctanceofSuperBowlticketholderstosellevenataveryhighprice?The
distinctivefeatureisthatboththeshoesthemerchantsellsyouandthemoneyyouspend
fromyourbudgetforshoesareheld“forexchange.”Theyareintendedtobetradedfor
othergoods.Othergoods,suchaswineandSuperBowltickets,areheld“foruse,”tobe
consumed or otherwise enjoyed. Your leisure time and the standard of living that your
incomesupportsarealsonotintendedforsaleorexchange.
Knetsch, Thaler, and I set out to design an experiment that would highlight the
contrastbetweengoodsthatareheldforuseandforexchange.Weborrowedoneaspectof
thedesignofourexperimentfromVernonSmith,thefounderofexperimentaleconomics,
withwhomIwouldshareaNobelPrizemanyyearslater.Inthismethod,alimitednumber
oftokens are distributed to theparticipants ina “market.” Any participants whoown a
token at the end Bon s A end Bon of the experiment can redeem it for cash. The
redemption values differ for different individuals, to represent the fact that the goods
tradedinmarketsaremorevaluabletosomepeoplethantoothers.Thesametokenmaybe
worth$10toyouand$20tome,andanexchangeatanypricebetweenthesevalueswill
beadvantageoustobothofus.
Smithcreatedvividdemonstrationsofhowwellthebasicmechanismsofsupplyand
demandwork.Individualswouldmakesuccessivepublicofferstobuyorsellatoken,and
otherswouldrespondpubliclytotheoffer.Everyonewatchestheseexchangesandseesthe
price at which the tokens change hands. The results are as regular as those of a
demonstrationinphysics.Asinevitablyaswaterflowsdownhill,thosewhoownatoken
thatisoflittlevaluetothem(becausetheirredemptionvaluesarelow)endupsellingtheir
tokenataprofittosomeonewhovaluesitmore.Whentradingends,thetokensareinthe
handsofthosewhocangetthemostmoneyforthemfromtheexperimenter.Themagicof
themarketshasworked!Furthermore,economictheorycorrectlypredictsboth the final
priceatwhichthemarketwillsettleandthenumberoftokensthatwillchangehands.If
halftheparticipantsinthemarketwererandomlyassignedtokens,thetheorypredictsthat
halfofthetokenswillchangehands.
WeusedavariationonSmith’smethodforourexperiment.Eachsessionbeganwith
several rounds of trades for tokens, which perfectly replicated Smith’s finding. The
estimatednumberoftradeswastypicallyverycloseoridenticaltotheamountpredicted
by the standard theory. The tokens, of course, had value only because they could be
exchangedfortheexperimenter’scash;theyhadnovalueforuse.Thenweconducteda
similarmarketforanobjectthatweexpectedpeopletovalueforuse:anattractivecoffee
mug, decorated with the university insignia of wherever we were conducting the
experiments.Themugwasthenworthabout$6(andwouldbeworthaboutdoublethat
amounttoday).Mugsweredistributedrandomlytohalftheparticipants.TheSellershad
theirmuginfrontofthem,andtheBuyerswereinvitedtolookattheirneighbor’smug;all
indicatedthepriceatwhichtheywouldtrade.TheBuyershadtousetheirownmoneyto
acquireamug.Theresultsweredramatic:theaveragesellingpricewasaboutdoublethe
averagebuyingprice,andtheestimatednumberoftradeswaslessthanhalfofthenumber
predictedbystandardtheory.Themagicofthemarketdidnotworkforagoodthatthe
ownersexpectedtouse.
Weconductedaseriesofexperimentsusingvariantsofthesameprocedure,always
withthesameresults.MyfavoriteisoneinwhichweaddedtotheSellersandBuyersa
thirdgroup—Choosers.UnliketheBuyers,whohadtospendtheirownmoneytoacquire
thegood,theChooserscouldreceiveeitheramugorasumofmoney,andtheyindicated
theamountofmoneythatwasasdesirableasreceivingthegood.Theseweretheresults:
Sellers $7.12
Choosers $3.12
Buyers $2.87
ThegapbetweenSellersandChoosersisremarkable,becausetheyactuallyfacethesame
choice!IfyouareaSelleryoucangohomewitheitheramBonsAamBonugormoney,
andifyouareaChooseryouhaveexactlythesametwooptions.Thelong-termeffectsof
thedecisionareidenticalforthetwogroups.Theonlydifferenceisintheemotionofthe
moment.ThehighpricethatSellerssetreflectsthereluctancetogiveupanobjectthat
theyalreadyown,areluctancethatcanbeseeninbabieswhoholdonfiercelytoatoyand
show great agitation when it is taken away. Loss aversion is built into the automatic
evaluationsofSystem1.
BuyersandChooserssetsimilarcashvalues,althoughtheBuyershavetopayforthe
mug, which is free for the Choosers. This is what we would expect if Buyers do not
experiencespendingmoneyonthemugasaloss.Evidencefrombrainimagingconfirms
thedifference.Sellinggoodsthatonewouldnormallyuseactivatesregionsofthebrain
thatareassociatedwithdisgustandpain.Buyingalsoactivatestheseareas,butonlywhen
the prices are perceived as too high—when you feel that a seller is taking money that
exceedstheexchangevalue.Brainrecordingsalsoindicatethatbuyingatespeciallylow
pricesisapleasurableevent.
ThecashvaluethattheSellerssetonthemugisabitmorethantwiceashighasthe
valuesetbyChoosersandBuyers.Theratioisveryclosetothelossaversioncoefficient
in risky choice, as we might expect if the same value function for gains and losses of
moneyisappliedtobothrisklessandriskydecisions.Aratioofabout2:1hasappearedin
studies of diverse economic domains, including the response of households to price
changes.Aseconomistswouldpredict,customerstendtoincreasetheirpurchasesofeggs,
orange juice, or fish when prices drop and to reduce their purchases when prices rise;
however, incontrast to thepredictionsof economictheory, theeffectofpriceincreases
(lossesrelativetothereferenceprice)isabouttwiceaslargeastheeffectofgains.
The mugs experiment has remained the standard demonstration of the endowment
effect, along with an even simpler experiment that Jack Knetsch reported at about the
sametime.Knetschaskedtwoclassestofilloutaquestionnaireandrewardedthemwitha
giftthatremainedinfrontofthemforthedurationoftheexperiment.Inonesession,the
prizewasanexpensivepen;inanother,abarofSwisschocolate.Attheendoftheclass,
theexperimentershowedthealternativegiftandallowedeveryonetotradehisorhergift
foranother.Onlyabout10%oftheparticipantsoptedtoexchangetheirgift.Mostofthose
whohadreceivedthepenstayedwiththepen,andthosewhohadreceivedthechocolate
didnotbudgeeither.
ThinkingLikeaTrader
Thefundamentalideasofprospecttheoryarethatreferencepointsexist,andthatlosses
loom larger than corresponding gains. Observations in real markets collected over the
yearsillustratethepoweroftheseconcepts.Astudyofthemarketforcondoapartmentsin
Boston during a downturn yielded particularly clear results. The authors of that study
compared the behavior of owners of similar units who had bought their dwellings at
different prices. For a rational agent, the buying price is irrelevant history—the current
marketvalueisallthatmatters.NotsoforHumansinadownmarketforhousing.Owners
whohave a high reference pointand thus face higher lossesset ahigher price on their
dwelling, spend a longer time trying to sell their home, and eventually receive more
money.
Theoriginaldemonstrationofanasymmetrybetweensellingpricesandbuyingprices
(or, more convincingly, between selling and choosing) was very important in the initial
acceptanceoftheideasofreferencepointandlossaversiBonsAersiBonon.However,it
is well understood that reference points are labile, especially in unusual laboratory
situations, and that the endowment effect can be eliminated by changing the reference
point.
Noendowmenteffectisexpectedwhenownersviewtheirgoodsascarriersofvalue
forfutureexchanges,awidespreadattitudeinroutinecommerceandinfinancialmarkets.
The experimental economist John List, who has studied trading at baseball card
conventions,foundthatnovicetraderswerereluctanttopartwiththecardstheyowned,
butthatthisreluctanceeventuallydisappearedwithtradingexperience.Moresurprisingly,
Listfoundalargeeffectoftradingexperienceontheendowmenteffectfornewgoods.
At a convention, List displayed a notice that invited people to take part in a short
survey, for which they would be compensated with a small gift: a coffee mug or a
chocolatebarofequalvalue.Thegiftswereassignedatrandom.Asthevolunteerswere
abouttoleave,Listsaidtoeachofthem,“Wegaveyouamug[orchocolatebar],butyou
cantradeforachocolatebar[ormug]instead,ifyouwish.”InanexactreplicationofJack
Knetsch’searlierexperiment,Listfoundthatonly18%oftheinexperiencedtraderswere
willingtoexchangetheirgiftfortheother.Insharpcontrast,experiencedtradersshowed
notraceofanendowmenteffect:48%ofthemtraded!Atleastinamarketenvironmentin
whichtradingwasthenorm,theyshowednoreluctancetotrade.
Jack Knetsch also conducted experiments in which subtle manipulations made the
endowmenteffectdisappear.Participantsdisplayedanendowmenteffectonlyiftheyhad
physical possession of the good for a while before the possibility of trading it was
mentioned.EconomistsofthestandardpersuasionmightbetemptedtosaythatKnetsch
had spent too much time with psychologists, because his experimental manipulation
showedconcernforthevariablesthatsocialpsychologistsexpecttobeimportant.Indeed,
thedifferentmethodologicalconcernsofexperimentaleconomistsandpsychologistshave
beenmuchinevidenceintheongoingdebateabouttheendowmenteffect.
Veterantradershave apparentlylearnedtoask the correctquestion,whichis “How
muchdoIwanttohavethatmug,comparedwithotherthingsIcouldhaveinstead?”This
is the question that Econs ask, and with this question there is no endowment effect,
because the asymmetry between the pleasure of getting and the pain of giving up is
irrelevant.
Recentstudiesofthepsychologyof“decisionmakingunderpoverty”suggestthatthe
poorareanothergroupinwhichwedonotexpecttofindtheendowmenteffect.Being
poor, inprospecttheory,is living below one’sreferencepoint.Therearegoodsthatthe
poorneedandcannotafford,sotheyarealways“inthelosses.”Smallamountsofmoney
thattheyreceivearethereforeperceivedasareducedloss,notasagain.Themoneyhelps
oneclimbalittletowardthereferencepoint,butthepooralwaysremainonthesteeplimb
ofthevaluefunction.
Peoplewhoarepoorthinkliketraders,butthedynamicsarequitedifferent.Unlike
traders, the poor are not indifferent to the differences between gaining and giving up.
Theirproblemisthatalltheirchoicesarebetweenlosses.Moneythatisspentononegood
isthelossofanothergoodthatcouldhavebeenpurchasedinstead.Forthepoor,costsare
losses.
Weallknowpeopleforwhomspendingispainful,althoughtheyareobjectivelyquite
well-off. There may also be cultural differences in the attitude toward money, and
especiallytowardthespendingofmoneyonwhimsBonsAhimsBonandminorluxuries,
such as the purchase of a decorated mug. Such a difference may explain the large
discrepancybetweentheresultsofthe“mugsstudy”intheUnitedStatesandintheUK.
Buying and selling prices diverge substantially in experiments conducted in samples of
students of the United States, but the differences are much smaller among English
students.Muchremainstobelearnedabouttheendowmenteffect.
SpeakingOfTheEndowmentEffect
“She didn’t care which of the two offices she would get, but a day after the
announcementwasmade,shewasnolongerwillingtotrade.Endowmenteffect!”
“Thesenegotiationsaregoingnowherebecausebothsidesfinditdifficulttomake
concessions, even when they can get something in return. Losses loom larger than
gains.”
“Whentheyraisedtheirprices,demanddriedup.”
“Hejusthatestheideaofsellinghishouseforlessmoneythanhepaidforit.Loss
aversionisatwork.”
“Heisamiser,andtreatsanydollarhespendsasaloss.”
P
BadEvents
Theconceptoflossaversioniscertainlythemostsignificantcontributionofpsychologyto
behavioraleconomics.Thisisodd,becausetheideathatpeopleevaluatemanyoutcomes
asgainsandlosses,andthatlossesloomlargerthangains,surprisesnoone.AmosandI
often joked that we were engaged in studying a subject about which our grandmothers
knewagreatdeal.Infact,however,weknowmorethanourgrandmothersdidandcan
nowembedlossaversioninthecontextofabroadertwo-systemsmodelofthemind,and
specificallyabiologicalandpsychologicalviewinwhichnegativityandescapedominate
positivity and approach. We can also trace the consequences of loss aversion in
surprisinglydiverseobservations:onlyout-of-pocketlossesarecompensatedwhengoods
are lost in transport; attempts at large-scale reforms very often fail; and professional
golfersputtmoreaccuratelyforparthanforabirdie.Cleverasshewas,mygrandmother
wouldhavebeensurprisedbythespecificpredictionsfromageneralideasheconsidered
obvious.
NegativityDominance
Figure12
Your heartbeat accelerated when you looked at the left-hand figure. It accelerated even
beforeyoucouldlabelwhatissoeerieaboutthatpicture.Aftersometimeyoumayhave
recognized the eyes of a terrified person. The eyes on the right, narrowed by the Crro
raisedcheeksofasmile,expresshappiness—andtheyarenotnearlyasexciting.Thetwo
pictureswerepresentedtopeoplelyinginabrainscanner.Eachpicturewasshownforless
than2/100ofasecondandimmediatelymaskedby“visualnoise,”arandomdisplayof
darkandbrightsquares.Noneofthe observers ever consciouslyknewthathehadseen
picturesofeyes,but one partoftheirbrain evidently knew: theamygdala,whichhas a
primary role as the “threat center” of the brain, although it is also activated in other
emotional states. Images of the brain showed an intense response of the amygdala to a
threatening picture that the viewer did not recognize. The information about the threat
probablytraveledviaasuperfastneuralchannelthatfeedsdirectlyintoapartofthebrain
that processes emotions, bypassing the visual cortex that supports the conscious
experienceof “seeing.”Thesame circuitalso causesschematicangry faces(a potential
threat) to be processed faster and more efficiently than schematic happy faces. Some
experimentershavereportedthatanangryface“popsout”ofacrowdofhappyfaces,but
asinglehappyfacedoesnotstandoutinanangrycrowd.Thebrainsofhumansandother
animalscontainamechanismthatisdesignedtogiveprioritytobadnews.Byshavinga
few hundredths of a second from the time needed to detect a predator, this circuit
improvestheanimal’soddsoflivinglongenoughtoreproduce.Theautomaticoperations
of System 1 reflect this evolutionary history. No comparably rapid mechanism for
recognizing good news has been detected. Of course, we and our animal cousins are
quickly alerted to signs of opportunities to mate or to feed, and advertisers design
billboardsaccordingly.Still,threatsareprivilegedaboveopportunities,astheyshouldbe.
The brain responds quickly even to purely symbolic threats. Emotionally loaded
wordsquicklyattractattention,andbadwords(war,crime)attractattentionfasterthando
happywords(peace,love).Thereisnorealthreat,butthemerereminderofabadeventis
treatedinSystem1asthreatening.Aswesawearlierwiththewordvomit,thesymbolic
representationassociativelyevokes inattenuated formmany of thereactions tothe real
thing,includingphysiologicalindicesofemotionandevenfractionaltendenciestoavoid
orapproach,recoilorleanforward.Thesensitivitytothreatsextendstotheprocessingof
statementsofopinionswithwhichwestronglydisagree.Forexample,dependingonyour
attitudetoeuthanasia,itwouldtakeyourbrainlessthanone-quarterofasecondtoregister
the “threat” in a sentence that starts with “I think euthanasia is an
acceptable/unacceptable…”
ThepsychologistPaulRozin,anexpertondisgust,observedthatasinglecockroach
willcompletelywrecktheappealofabowlofcherries,butacherrywilldonothingatall
for a bowl of cockroaches. As he points out, the negative trumps the positive in many
ways,andlossaversionisoneofmanymanifestationsofabroadnegativitydominance.
Otherscholars,inapapertitled“BadIsStrongerThanGood,”summarizedtheevidence
asfollows:“Bademotions,bad parents, and badfeedbackhavemoreimpact than good
ones, and bad information is processed more thoroughly than good. The self is more
motivatedtoavoidbadself-definitionsthantopursuegoodones.Badimpressionsandbad
stereotypes are quicker to form and more resistant to disconfirmation than good ones.”
TheyciteJohnGottman,thewell-knownexpertinmaritalrelations,whoobservedthatthe
long-term success of a relationship depends far more on avoiding the negative than on
seeking the positive. Gottman estimated that a stable relationship requires Brro Qres
Brrthatgoodinteractionsoutnumberbadinteractionsbyatleast5to1.Otherasymmetries
inthesocialdomainareevenmorestriking.Weallknowthatafriendshipthatmaytake
yearstodevelopcanberuinedbyasingleaction.
Somedistinctionsbetweengoodandbadarehardwiredintoourbiology.Infantsenter
theworldreadytorespondtopainasbadandtosweet(uptoapoint)asgood.Inmany
situations,however,theboundarybetweengoodandbadisareferencepointthatchanges
overtimeanddependsontheimmediatecircumstances.Imaginethatyouareoutinthe
countryonacoldnight,inadequatelydressedforthetorrentialrain,yourclothessoaked.A
stingingcoldwindcompletesyourmisery.Asyouwanderaround,youfindalargerock
thatprovidessomeshelterfromthefuryoftheelements.ThebiologistMichelCabanac
would call the experience of that moment intensely pleasurable because it functions, as
pleasurenormallydoes,toindicatethedirectionofabiologicallysignificantimprovement
ofcircumstances.Thepleasantreliefwillnotlastverylong,ofcourse,andyouwillsoon
beshiveringbehindtherockagain,drivenbyyourrenewedsufferingtoseekbettershelter.
GoalsareReferencePoints
Lossaversionreferstotherelativestrengthoftwomotives:wearedrivenmorestronglyto
avoidlossesthantoachievegains.Areferencepointissometimesthestatusquo,butit
canalsobeagoalinthefuture:notachievingagoalisaloss,exceedingthegoalisagain.
Aswemightexpectfromnegativitydominance,thetwomotivesarenotequallypowerful.
Theaversion to the failure ofnot reachingthe goal ismuch strongerthan thedesire to
exceedit.
Peopleoftenadoptshort-termgoalsthattheystrivetoachievebutnotnecessarilyto
exceed.Theyarelikelytoreducetheireffortswhentheyhavereachedanimmediategoal,
with results that sometimes violate economic logic. New York cabdrivers, for example,
mayhaveatargetincomeforthemonthortheyear,butthegoalthatcontrolstheireffortis
typicallyadailytargetofearnings.Ofcourse,thedailygoalismucheasiertoachieve(and
exceed)onsomedaysthanonothers.Onrainydays,aNewYorkcabneverremainsfree
forlong,andthedriverquicklyachieveshistarget;notsoinpleasantweather,whencabs
often waste time cruising the streets looking for fares. Economic logic implies that
cabdriversshouldworkmanyhoursonrainydaysandtreatthemselvestosomeleisureon
mild days, when they can “buy” leisure at a lower price. The logic of loss aversion
suggeststheopposite:driverswhohaveafixeddailytargetwillworkmanymorehours
whenthepickingsareslimandgohomeearlywhenrain-drenchedcustomersarebegging
tobetakensomewhere.
The economists Devin Pope and Maurice Schweitzer, at the University of
Pennsylvania, reasoned that golf provides a perfect example of a reference point: par.
Everyholeonthegolfcoursehasanumberofstrokesassociatedwithit;theparnumber
provides the baseline for good—but not outstanding—performance. For a professional
golfer,abirdie(onestrokeunderpar)isagain,andabogey(onestrokeoverpar)isaloss.
Theeconomistscomparedtwosituationsaplayermightfacewhennearthehole:
putttoavoidabogey
putttoachieveabirdie
Everystrokecountsingolf,andinprofessionalgolfeverystrokecountsalot.According
toprospecttheory,however,somestrokescountmorethanothers.Failingtomakeparisa
los Brro Q los Brrs, but missing a birdie putt is a foregone gain, not a loss. Pope and
Schweitzerreasonedfromlossaversionthatplayerswouldtryalittleharderwhenputting
forpar(toavoidabogey)thanwhenputtingforabirdie.Theyanalyzedmorethan2.5
millionputtsinexquisitedetailtotestthatprediction.
Theywereright.Whethertheputtwaseasyorhard,ateverydistancefromthehole,
theplayersweremoresuccessfulwhenputtingforparthanforabirdie.Thedifferencein
theirrateofsuccesswhengoingforpar(toavoidabogey)orforabirdiewas3.6%.This
differenceisnottrivial.TigerWoodswasoneofthe“participants”intheirstudy.Ifinhis
best years Tiger Woods had managed to putt as well for birdies as he did for par, his
averagetournamentscorewouldhaveimprovedbyonestrokeandhisearningsbyalmost
$1 million per season. These fierce competitors certainly do not make a conscious
decision to slack off on birdie putts, but their intense aversion to a bogey apparently
contributestoextraconcentrationonthetaskathand.
Thestudyofputtsillustratesthepowerofatheoreticalconceptasanaidtothinking.
Who would have thought it worthwhile to spend months analyzing putts for par and
birdie?Theideaoflossaversion,whichsurprisesnooneexceptperhapssomeeconomists,
generated a precise and nonintuitive hypothesis and led researchers to a finding that
surprisedeveryone—includingprofessionalgolfers.
DefendingtheStatusQuo
Ifyouaresettolookforit,theasymmetricintensityofthemotivestoavoidlossesandto
achievegainsshowsupalmosteverywhere.Itisanever-presentfeatureofnegotiations,
especially of renegotiations of an existing contract, the typical situation in labor
negotiations and in international discussions of trade or arms limitations. The existing
termsdefinereferencepoints, andaproposed change inanyaspect of theagreementis
inevitablyviewedasaconcessionthatonesidemakestotheother.Lossaversioncreates
anasymmetrythatmakesagreementsdifficulttoreach.Theconcessionsyoumaketome
aremygains,buttheyareyourlosses;theycauseyoumuchmorepainthantheygiveme
pleasure.Inevitably,youwillplaceahighervalueonthemthanIdo.Thesameistrue,of
course,oftheverypainfulconcessionsyoudemandfromme,whichyoudonotappearto
valuesufficiently!Negotiationsoverashrinkingpieareespeciallydifficult,becausethey
requireanallocationoflosses.Peopletendtobemuchmoreeasygoingwhentheybargain
overanexpandingpie.
Many of the messages that negotiators exchange in the course of bargaining are
attemptstocommunicateareferencepointandprovideananchortotheotherside.The
messages are not always sincere. Negotiators often pretend intense attachment to some
good(perhapsmissilesofaparticulartypeinbargainingoverarmsreductions),although
theyactuallyviewthatgoodasabargainingchipandintendultimatelytogiveitawayin
anexchange.Becausenegotiatorsareinfluencedbyanormofreciprocity, a concession
thatispresentedaspainfulcallsforanequallypainful(andperhapsequallyinauthentic)
concessionfromtheotherside.
Animals,includingpeople,fighthardertopreventlossesthantoachievegains.Inthe
worldofterritorialanimals,thisprincipleexplainsthesuccessofdefenders.Abiologist
observedthat“whenaterritoryholderischallengedbyarival,theowneralmostalways
winsthecontest—usuallywithinamatterofseconds.”Inhumanaffairs,thesamesimple
rule explains much of what happens when institutions attempt to reform themselves, in
“reo Brro Q;reo Brrrganizations” and “restructuring” of companies, and in efforts to
rationalize a bureaucracy, simplify the tax code, or reduce medical costs. As initially
conceived,plansforreformalmostalwaysproducemanywinnersandsomeloserswhile
achieving an overall improvement. If the affected parties have any political influence,
however,potentialloserswillbemoreactiveanddeterminedthanpotentialwinners;the
outcomewillbebiasedintheirfavorandinevitablymoreexpensiveandlesseffectivethan
initially planned. Reforms commonly include grandfather clauses that protect current
stake-holders—for example, when the existing workforce is reduced by attrition rather
thanbydismissals,orwhencutsinsalariesandbenefitsapplyonlytofutureworkers.Loss
aversionisapowerfulconservativeforcethatfavorsminimalchangesfromthestatusquo
inthelivesofbothinstitutionsandindividuals.Thisconservatismhelpskeepusstablein
ourneighborhood,ourmarriage,andourjob;itisthegravitationalforcethatholdsourlife
togethernearthereferencepoint.
LossAversionintheLaw
During the year that we spent working together in Vancouver, Richard Thaler, Jack
Knetsch, and I were drawn into a study of fairness in economic transactions, partly
becausewewereinterestedinthetopicbutalsobecausewehadanopportunityaswellas
an obligation to make up a new questionnaire every week. The Canadian government’s
Department of Fisheries and Oceans had a program for unemployed professionals in
Toronto,whowerepaidtoadministertelephonesurveys.Thelargeteamofinterviewers
worked every night and new questions were constantly needed to keep the operation
going.ThroughJackKnetsch,weagreedtogenerateaquestionnaireeveryweek,infour
color-labeled versions. We could ask about anything; the only constraint was that the
questionnaire should include at least one mention of fish, to make it pertinent to the
missionofthedepartment.Thiswentonformanymonths,andwetreatedourselvestoan
orgyofdatacollection.
We studied public perceptions of what constitutes unfair behavior on the part of
merchants, employers, and landlords. Our overarching question was whether the
opprobriumattachedtounfairnessimposesconstraintsonprofitseeking.Wefoundthatit
does.Wealsofoundthatthemoralrulesbywhichthepublicevaluateswhatfirmsmayor
maynotdodrawacrucialdistinctionbetweenlossesandgains.Thebasicprincipleisthat
the existing wage, price, or rent sets a reference point, which has the nature of an
entitlementthatmustnotbeinfringed.Itisconsideredunfairforthefirmtoimposelosses
onitscustomersorworkersrelativetothereferencetransaction,unlessitmustdosoto
protectitsownentitlement.Considerthisexample:
Ahardwarestorehasbeensellingsnowshovelsfor$15.Themorningafteralarge
snowstorm,thestoreraisesthepriceto$20.
Pleaseratethisactionas:
CompletelyFairAcceptableUnfairVeryUnfair
Thehardwarestorebehavesappropriatelyaccordingtothestandardeconomicmodel:it
respondstoincreaseddemandbyraisingitsprice.Theparticipantsinthesurveydidnot
agree:82%ratedtheactionUnfairorVeryUnfair.Theyevidentlyviewedthepre-blizzard
price as a reference point and the raised price as a loss that the store imposes on its
customers, not because it must but simply because it can. A basic rule of fairness, we
found,iBrroQd,iBrrsthattheexploitationofmarketpowertoimposelossesonothersis
unacceptable. The following example illustrates this rule in another context (the dollar
values should be adjusted for about 100% inflation since these data were collected in
1984):
Asmallphotocopyingshophasoneemployeewhohasworkedthereforsixmonths
andearns$9perhour.Businesscontinuestobesatisfactory,butafactoryinthearea
has closed and unemployment has increased. Other small shops have now hired
reliableworkersat$7anhourtoperformjobssimilartothosedonebythephotocopy
shopemployee.Theowneroftheshopreducestheemployee’swageto$7.
The respondents did not approve: 83% considered the behavior Unfair or Very Unfair.
However, a slight variation on the question clarifies the nature of the employer’s
obligation.Thebackgroundscenarioofaprofitablestoreinanareaofhighunemployment
isthesame,butnow
thecurrentemployeeleaves,andtheownerdecidestopayareplacement$7anhour.
A large majority (73%) considered this action Acceptable. It appears that the employer
does not have a moral obligation to pay $9 an hour. The entitlement is personal: the
currentworkerhasarighttoretainhiswageevenifmarketconditionswouldallowthe
employer to impose a wage cut. The replacement worker has no entitlement to the
previousworker’sreference wage, andthe employeris thereforeallowedto reducepay
withouttheriskofbeingbrandedunfair.
The firm has its own entitlement, which is to retain its current profit. If it faces a
threat of a loss, it is allowed to transfer the loss to others. A substantial majority of
respondentsbelievedthatitisnotunfairforafirmtoreduceitsworkers’wageswhenits
profitabilityisfalling.Wedescribedtherulesasdefiningdualentitlementstothefirmand
toindividualswithwhomitinteracts.Whenthreatened,itisnotunfairforthefirmtobe
selfish.Itisnotevenexpectedtotakeonpartofthelosses;itcanpassthemon.
Different rules governed what the firm could do to improve its profits or to avoid
reducedprofits.Whenafirmfacedlowerproductioncosts,therulesoffairnessdidnot
require it to share the bonanza with either its customers or its workers. Of course, our
respondentslikedafirmbetteranddescribeditasmorefairifitwasgenerouswhenits
profitsincreased,buttheydidnotbrandasunfairafirmthatdidnotshare.Theyshowed
indignationonlywhenafirmexploiteditspowertobreakinformalcontractswithworkers
orcustomers,andtoimposealossonothersinordertoincreaseitsprofit.Theimportant
taskforstudentsofeconomicfairnessisnottoidentifyidealbehaviorbuttofindtheline
thatseparatesacceptableconductfromactionsthatinviteopprobriumandpunishment.
WewerenotoptimisticwhenwesubmittedourreportofthisresearchtotheAmerican
EconomicReview.Ourarticlechallengedwhatwasthenacceptedwisdomamongmany
economiststhateconomicbehaviorisruledbyself-interestandthatconcernsforfairness
are generally irrelevant. We also relied on the evidence of survey responses, for which
economistsgenerallyhavelittlerespect.However,theeditorofthejournalsentourarticle
for evaluation to two economists who were not bound by those conventions (we later
learnedtheiridentity;theywerethemostfriendlytheeditorcouldhavefound).Theeditor
madethecorrectcall.Thearticleisoftencited,anditsconclusionsBrroQionsBrrhave
stoodthetestoftime.Morerecentresearchhassupportedtheobservationsofreference-
dependentfairnessandhasalsoshownthatfairnessconcernsareeconomicallysignificant,
afact wehad suspectedbut didnot prove. Employers whoviolate rulesof fairnessare
punishedbyreducedproductivity,andmerchantswhofollowunfairpricingpoliciescan
expecttolosesales.Peoplewholearnedfromanewcatalogthatthemerchantwasnow
charginglessforaproductthattheyhadrecentlyboughtatahigherpricereducedtheir
future purchases from that supplier by 15%, an average loss of $90 per customer. The
customers evidently perceived the lower price as the reference point and thought of
themselves as having sustained a loss by paying more than appropriate. Moreover, the
customerswhoreactedthemoststronglywerethosewhoboughtmoreitemsandathigher
prices.Thelossesfar exceeded the gainsfromtheincreasedpurchasesproducedbythe
lowerpricesinthenewcatalog.
Unfairly imposing losses on people can be risky if the victims are in a position to
retaliate. Furthermore, experiments have shown that strangers who observe unfair
behavior often join in the punishment. Neuroeconomists (scientists who combine
economicswithbrainresearch)haveusedMRImachinestoexaminethebrainsofpeople
who are engaged in punishing one stranger for behaving unfairly to another stranger.
Remarkably,altruisticpunishmentisaccompaniedbyincreasedactivityinthe“pleasure
centers”ofthebrain.Itappearsthatmaintainingthesocialorderandtherulesoffairness
inthisfashionisitsownreward.Altruisticpunishmentcouldwellbethegluethatholds
societiestogether.However,ourbrainsarenotdesignedtorewardgenerosityasreliablyas
theypunishmeanness.Hereagain,wefindamarkedasymmetrybetweenlossesandgains.
The influence of loss aversion and entitlements extends far beyond the realm of
financialtransactions.Juristswerequicktorecognizetheirimpactonthelawandinthe
administration of justice. In one study, David Cohen and Jack Knetsch found many
examples of a sharp distinction between actual losses and foregone gains in legal
decisions.Forexample,amerchantwhosegoodswerelostintransitmaybecompensated
for costs he actually incurred, but is unlikely to be compensated for lost profits. The
familiar rule that possession is nine-tenths of the law confirms the moral status of the
referencepoint.Inamorerecentdiscussion,EyalZamirmakestheprovocativepointthat
thedistinctiondrawninthelawbetweenrestoringlossesandcompensatingforforegone
gainsmay bejustifiedby theirasymmetrical effectson individual well-being.If people
wholosesuffermorethanpeoplewhomerelyfailtogain,theymayalsodeservemore
protectionfromthelaw.
SpeakingofLosses
“Thisreformwillnotpass.Thosewhostandtolosewillfightharderthanthosewho
standtogain.”
“Eachofthemthinkstheother’sconcessionsarelesspainful.Theyarebothwrong,
ofcourse.It’sjusttheasymmetryoflosses.”
“Theywouldfinditeasiertorenegotiatetheagreementiftheyrealizedthepiewas
actuallyexpanding.They’renotallocatinglosses;theyareallocatinggains.”
“Rental prices around here have gone up r Brro Qup r Brrecently, but our tenants
don’t think it’s fair that we should raise their rent, too. They feel entitled to their
currentterms.”
“Myclientsdon’tresentthepricehikebecausetheyknowmycostshavegoneup,
too.Theyacceptmyrighttostayprofitable.”
P
TheFourfoldPattern
Wheneveryouformaglobalevaluationofacomplexobject—acaryoumay buy,your
son-in-law, or an uncertain situation—you assign weights to its characteristics. This is
simplyacumbersomewayofsayingthatsomecharacteristicsinfluenceyourassessment
more than others do. The weighting occurs whether or not you are aware of it;it is an
operationofSystem1.Youroverallevaluationofacarmayputmoreorlessweighton
gas economy, comfort, or appearance. Your judgment of your son-in-law may depend
moreorlessonhowrichorhandsomeorreliableheis.Similarly,yourassessmentofan
uncertain prospect assigns weights to the possible outcomes. The weights are certainly
correlatedwiththeprobabilitiesoftheseoutcomes:a50%chancetowinamillionismuch
moreattractivethana1%chancetowinthesameamount.Theassignmentofweightsis
sometimesconsciousanddeliberate.Mostoften,however,youarejustanobservertoa
globalevaluationthatyourSystem1delivers.
ChangingChances
Onereasonforthepopularityofthegamblingmetaphorinthestudyofdecisionmakingis
thatitprovidesanaturalrulefortheassignmentofweightstotheoutcomesofaprospect:
themoreprobableanoutcome,themoreweightitshouldhave.Theexpectedvalueofa
gambleistheaverageofitsoutcomes,eachweightedbyitsprobability.Forexample,the
expectedvalueof“20%chancetowin$1,000and75%chancetowin$100”is$275.In
thepre-Bernoullidays,gambleswereassessedbytheirexpectedvalue.Bernoulliretained
this method for assigning weights to the outcomes, which is known as the expectation
principle, but applied it to the psychological value of the outcomes. The utility of a
gamble,inhistheory,istheaverageoftheutilitiesofitsoutcomes,eachweightedbyits
probability.
The expectation principle does not correctly describe how you think about the
probabilities related to risky prospects. In the four examples below, your chances of
receiving$1millionimproveby5%.Isthenewsequallygoodineachcase?
A.From0to5%
B.From5%to10%
C.From60%to65%
D.From95%to100%
Theexpectationprincipleassertsthatyourutilityincreasesineachcasebyexactly5%of
the utility of receiving $1 million. Does this prediction describe your experiences? Of
coursenot.
Everyoneagreesthat0 5%and95% 100%aremoreimpressivethaneither5%
10% or 60% 65%. Increasing the chances from 0 to 5% transforms the situation,
creating a possibility that did not exist earlier, a hope of winning the prize. It is a
qualitativechange,where5 10%isonlyaquantitativeimprovement.Thechangefrom
5% to 10% doubles the probability of winning, but there is general agreement that the
psychological value of the prospect does not double. The large impact of 0 5%
illustrates the possibility effect, which causes highly unlikely outcomes to be weighted
disproportionately more than they “deserve.” People who buy lottery tickets in vast
amountsshowthemselveswillingtopaymuchmorethanexpectedvalueforverysmall
chancestowinalargeprize.
Theimprovementfrom95%to100% isanotherqualitativechangethathasa large
impact,thecertaintyeffect.Outcomesthatarealmostcertainaregivenlessweightthan
theirprobabilityjustifies.Toappreciatethecertaintyeffect,imaginethatyouinherited$1
million,butyourgreedystepsisterhascontestedthewillincourt.Thedecisionisexpected
tomorrow.Yourlawyerassuresyouthatyouhaveastrongcaseandthatyouhavea95%
chancetowin,buthetakespainstoremindyouthatjudicialdecisionsareneverperfectly
predictable.Nowyouareapproachedbyarisk-adjustmentcompany,whichofferstobuy
yourcasefor$910,000outright—takeitorleaveit.Theofferislower(by$40,000!)than
theexpectedvalueofwaitingforthejudgment(whichis$950,000),butareyouquitesure
youwould wantto reject it?If suchan eventactuallyhappens inyour life,you should
knowthatalargeindustryof“structuredsettlements”existstoprovidecertaintyatahefty
price,bytakingadvantageofthecertaintyeffect.
Possibility and certainty have similarly powerful effects in the domain of losses.
Whenalovedoneiswheeledintosurgery,a5%riskthatanamputationwillbenecessary
isverybad—muchmorethanhalfasbadasa10%risk.Becauseofthepossibilityeffect,
wetendtooverweightsmallrisksandarewillingtopayfarmorethanexpectedvalueto
eliminate them altogether. The psychological difference between a 95% risk of disaster
andthecertaintyofdisasterappearstobeevengreater;thesliverofhopethateverything
couldstillbeokayloomsverylarge.Overweightingofsmallprobabilitiesincreasesthe
attractivenessofbothgamblesandinsurancepolicies.
The conclusion is straightforward: the decision weights that people assign to
outcomes are not identical to the probabilities of these outcomes, contrary to the
expectation principle. Improbable outcomes are overweighted—this is the possibility
effect.Outcomesthatarealmostcertainareunderweightedrelativetoactualcertainty.The
expectation principle, by which values are weighted by their probability, is poor
psychology.
The plot thickens, however, because there is a powerful argument that a decision
makerwhowishestoberationalmustconformtotheexpectationprinciple.Thiswasthe
mainpointoftheaxiomaticversionofutilitytheorythatvonNeumannandMorgenstern
introduced in 1944. They proved that any weighting of uncertain outcomes that is not
strictly proportional to probability leads to inconsistencies and other disasters. Their
derivation of the expectation principle from axioms of rational choice was immediately
recognizedasamonumentalachievement,whichplacedexpectedutilitytheoryatthecore
of the rational agent model in economics and other social sciences. Thirty years later,
when Amos introduced me to their work, he presented it as an object of awe. He also
introducedmeBimaameBimtoafamouschallengetothattheory.
Allais’sParadox
In1952,afewyearsafterthepublicationofvonNeumannandMorgenstern’stheory,a
meeting was convened in Paris to discuss the economics of risk. Many of the most
renownedeconomistsofthetimewereinattendance.TheAmericanguestsincludedthe
futureNobellaureatesPaulSamuelson,KennethArrow,andMiltonFriedman,aswellas
theleadingstatisticianJimmieSavage.
One of the organizers of the Paris meeting was Maurice Allais, who would also
receiveaNobelPrizesomeyearslater.Allaishadsomethinguphissleeve,acoupleof
questionsonchoicethathepresentedtohisdistinguishedaudience.Inthetermsofthis
chapter,Allaisintendedtoshowthathisguestsweresusceptibletoacertaintyeffectand
thereforeviolatedexpectedutilitytheoryandtheaxiomsofrationalchoiceonwhichthat
theoryrests.ThefollowingsetofchoicesisasimplifiedversionofthepuzzlethatAllais
constructed.InproblemsAandB,whichwouldyouchoose?
A.61%chancetowin$520,000OR63%chancetowin$500,000
B.98%chancetowin$520,000OR100%chancetowin$500,000
Ifyouarelikemostotherpeople,youpreferredtheleft-handoptioninproblemAandyou
preferredtheright-handoptioninproblemB.Ifthesewereyourpreferences,youhavejust
committed a logical sin and violated the rules of rational choice. The illustrious
economistsassembledinPariscommittedsimilarsinsinamoreinvolvedversionofthe
“Allaisparadox.”
To see why these choices are problematic, imagine that the outcome will be
determinedbyablinddrawfromanurnthatcontains100marbles—youwinifyoudrawa
redmarble,youloseifyoudrawwhite.InproblemA,almosteverybodypreferstheleft-
handurn,althoughithasfewerwinningredmarbles,becausethedifferenceinthesizeof
theprizeismoreimpressivethanthedifferenceinthechancesofwinning.InproblemB,a
largemajoritychoosestheurnthatguaranteesagainof$500,000.Furthermore,peopleare
comfortablewithbothchoices—untiltheyareledthroughthelogicoftheproblem.
Comparethetwoproblems,andyouwillseethatthetwournsofproblemBaremore
favorable versions of the urns of problem A, with 37 white marbles replaced by red
winning marbles in each urn. The improvement on the left is clearly superior to the
improvementontheright,sinceeachredmarblegivesyouachancetowin$520,000on
the left and only $500,000 on the right. So you started in the first problem with a
preferencefortheleft-handurn,whichwasthenimprovedmorethantheright-handurn—
butnowyouliketheoneontheright!Thispatternofchoicesdoesnotmakelogicalsense,
butapsychologicalexplanationisreadilyavailable:thecertaintyeffectisatwork.The2%
difference between a 100% and a 98% chance to win in problem B is vastly more
impressivethanthesamedifferencebetween63%and61%inproblemA.
AsAllaishadanticipated,thesophisticatedparticipantsatthemeetingdidnotnotice
thattheirpreferencesviolatedutilitytheoryuntilhedrewtheirattentiontothatfactasthe
meetingwasabouttoend.Allaishadintendedthisannouncementtobeabombshell:the
leadingdecisiontheoristsin the worldhadpreferencesthat wereinconsistentwiththeir
ownviewofrationality!Heapparentlybelievedthathisaudiencewouldbepersuadedto
giveuptheapproachthatBimaahatBimherathercontemptuouslylabeled“theAmerican
school” and adopt an alternative logic of choice that he had developed. He was to be
sorelydisappointed.
Economists who were not aficionados of decision theory mostly ignored the Allais
problem.Asoftenhappenswhenatheorythathasbeenwidelyadoptedandfounduseful
ischallenged,theynotedtheproblemasananomalyandcontinuedusingexpectedutility
theoryasifnothinghadhappened.Incontrast,decisiontheorists—amixedcollectionof
statisticians, economists, philosophers, and psychologists—took Allais’s challenge very
seriously. When AmosandI beganourwork, oneofour initialgoalswasto developa
satisfactorypsychologicalaccountofAllais’sparadox.
Most decision theorists, notably including Allais, maintained their belief in human
rationality and tried to bend the rules of rational choice to make the Allais pattern
permissible. Over the years there have been multiple attempts to find a plausible
justification for the certainty effect, none very convincing. Amos had little patience for
these efforts; he called the theorists who tried to rationalize violations of utility theory
“lawyersforthemisguided.”Wewentinanotherdirection.Weretainedutilitytheoryasa
logicofrationalchoicebutabandonedtheideathatpeopleareperfectlyrationalchoosers.
Wetookonthetaskofdevelopingapsychologicaltheorythatwoulddescribethechoices
peoplemake,regardlessofwhethertheyarerational.Inprospecttheory,decisionweights
wouldnotbeidenticaltoprobabilities.
DecisionWeights
Manyyearsafterwepublishedprospecttheory,AmosandIcarriedoutastudyinwhich
wemeasuredthedecisionweightsthatexplainedpeople’spreferencesfor gambles with
modestmonetarystakes.Theestimatesforgainsareshownintable4.
Table4
Youcanseethatthedecisionweightsareidenticaltothecorrespondingprobabilitiesatthe
extremes:bothequalto0whentheoutcomeisimpossible,andbothequalto100whenthe
outcomeisasurething.However,decisionweightsdepartsharplyfromprobabilitiesnear
these points. At the low end, we find the possibility effect: unlikely events are
considerably overweighted. For example, the decision weight that corresponds to a 2%
chanceis8.1.Ifpeopleconformedtotheaxiomsofrationalchoice,thedecisionweight
wouldbe2—sotherareeventisoverweightedbyafactorof4.Thecertaintyeffectatthe
otherendoftheprobabilityscaleisevenmorestriking.A2%riskofnotwinningtheprize
reducestheutilityofthegambleby13%,from100to87.1.
Toappreciate the asymmetrybetweenthe possibilityeffectand thecertaintyeffect,
imaginefirstthatyouhavea1%chancetowin$1million.Youwillknowtheoutcome
tomorrow.Now,imaginethatyouarealmostcertaintowin$1million,butthereisa1%
chancethatyouwillnot.Again,youwilllearntheoutcometomorrow.Theanxietyofthe
secondsituationappearstobemoresalientthanthehopeinthefirst.Thecertaintyeffect
isalsomorestrikingthanthepossibilityeffectiftheoutcomeisasurgicaldisasterrather
thanafinancialgain.Comparetheintensitywithwhichyoufocusonthefaintsliverof
hopeinanoperationthatisalmostcertaintobefatal,comparedtothefearofa1%risk.
<Bimaav><Bimpheight=“0%”width=“5%”>Thecombinationofthecertaintyeffect
andpossibilityeffectsatthetwoendsoftheprobabilityscaleisinevitablyaccompanied
by inadequate sensitivity to intermediate probabilities. You can see that the range of
probabilitiesbetween5% and95%is associatedwitha muchsmallerrange ofdecision
weights (from 13.2 to 79.3), about two-thirds as much as rationally expected.
Neuroscientists have confirmed these observations, finding regions of the brain that
respondtochangesintheprobabilityofwinningaprize.Thebrain’sresponsetovariations
ofprobabilitiesisstrikinglysimilartothedecisionweightsestimatedfromchoices.
Probabilitiesthatareextremelyloworhigh(below1%orabove99%)areaspecial
case.Itisdifficulttoassignauniquedecisionweighttoveryrareevents,becausetheyare
sometimesignoredaltogether,effectivelyassignedadecisionweightofzero.Ontheother
hand,whenyoudonotignoretheveryrareevents,youwillcertainlyoverweightthem.
Mostofusspendverylittletimeworryingaboutnuclearmeltdownsorfantasizingabout
largeinheritancesfromunknownrelatives.However,whenanunlikelyeventbecomesthe
focus of attention, we will assign it much more weight than its probability deserves.
Furthermore,peoplearealmostcompletelyinsensitivetovariationsofriskamongsmall
probabilities.Acancerriskof0.001%isnoteasilydistinguishedfromariskof0.00001%,
although the former would translate to 3,000 cancers for the population of the United
States,andthelatterto30.
When you pay attention to a threat, you worry—and the decision weights reflect how
much you worry. Because of the possibility effect, the worry is not proportional to the
probabilityofthethreat.Reducingormitigatingtheriskisnotadequate;toeliminatethe
worrytheprobabilitymustbebroughtdowntozero.
Thequestionbelowisadaptedfromastudyoftherationalityofconsumervaluations
ofhealthrisks,whichwaspublishedbyateamofeconomistsinthe1980s.Thesurvey
wasaddressedtoparentsofsmallchildren.
Supposethatyoucurrentlyuseaninsectspraythatcostsyou$10perbottleandit
resultsin15inhalationpoisoningsand15childpoisoningsforevery10,000bottles
ofinsectspraythatareused.
Youlearnofamoreexpensiveinsecticidethatreduceseachoftherisksto5forevery
10,000bottles.Howmuchwouldyoubewillingtopayforit?
Theparentswerewillingto payan additional$2.38,onaverage, to reducetherisksby
two-thirds from 15 per 10,000 bottles to 5. They were willing to pay $8.09, more than
threetimesasmuch,toeliminateitcompletely.Otherquestionsshowedthattheparents
treatedthetworisks(inhalationandchildpoisoning)asseparateworriesandwerewilling
topaya certaintypremiumforthecompleteeliminationof eitherone.Thispremiumis
compatiblewiththepsychologyofworrybutnotwiththerationalmodel.
TheFourfoldPattern
When Amos and I began our work on prospect theory, we quickly reached two
conclusions: people attach values to gains and losses rather than to wealth, and the
decisionweightsthattheyassigntooutcomesaredifferentfromprobabilities.Neitheridea
wascompletelynew,butincombinationtheyexplainedadistinctivepatternofpreferences
thatwecaBimaaecaBimlledthefourfoldpattern.Thenamehasstuck.Thescenariosare
illustratedbelow.
Figure13
Thetoprowineachcellshowsanillustrativeprospect.
Thesecondrowcharacterizesthefocalemotionthattheprospectevokes.
Thethirdrow indicateshowmostpeoplebehavewhenofferedachoicebetweena
gambleandasuregain(orloss)thatcorrespondstoitsexpectedvalue(forexample,
between“95%chancetowin$10,000”and“$9,500withcertainty”).Choicesaresaid
toberiskaverseifthesurethingispreferred,riskseekingifthegambleispreferred.
Thefourthrowdescribestheexpectedattitudesofadefendantandaplaintiffasthey
discussasettlementofacivilsuit.
Thefourfoldpatternofpreferencesisconsideredoneofthecoreachievementsofprospect
theory.Threeofthefourcellsarefamiliar;thefourth(topright)wasnewandunexpected.
ThetopleftistheonethatBernoullidiscussed:peopleareaversetoriskwhenthey
considerprospectswithasubstantialchancetoachievealargegain.Theyarewilling
toacceptlessthantheexpectedvalueofagambletolockinasuregain.
Thepossibilityeffectinthebottomleftcellexplainswhylotteriesarepopular.When
thetopprizeisverylarge,ticketbuyersappearindifferenttothefactthattheirchance
of winning is minuscule. A lottery ticket is the ultimate example of the possibility
effect.Withoutaticketyoucannotwin,withaticketyouhaveachance,andwhether
thechanceistinyormerelysmallmatterslittle.Ofcourse,whatpeopleacquirewith
aticketismorethanachancetowin;itistherighttodreampleasantlyofwinning.
Thebottomrightcelliswhereinsuranceisbought.Peoplearewillingtopaymuch
moreforinsurancethan expected value—whichishowinsurancecompanies cover
their costs and make their profits. Here again, people buy more than protection
againstanunlikelydisaster;theyeliminateaworryandpurchasepeaceofmind.
The results for the top right cell initially surprised us. We were accustomed to think in
termsofriskaversionexceptforthebottomleftcell,wherelotteriesarepreferred.When
welookedatourchoicesfor badoptions,wequicklyrealizedthatwewere just asrisk
seekinginthedomainoflossesaswewereriskaverseinthedomainofgains.Wewere
not the first to observe risk seeking with negative prospects—at least two authors had
reportedthatfact,buttheyhadnotmademuchofit.However,wewerefortunatetohavea
frameworkthatmadethefindingofriskseekingeasytointerpret,andthatwasamilestone
inourthinking.Indeed,weidentifiedtworeasonsforthiseffect.
First, there is diminishing sensitivity. The sure loss is very aversive because the
reactiontoalossof$900ismorethan90%asintenseasthereactiontoalossof$1,000.
Thesecondfactormaybeevenmorepowerful:thedecisionweightthatcorrespondstoa
probabilityof90%isonlyabout71,muchlowerthantheprobability.Theresultisthat
whenyouconsiderachoicebetweenasurelossandagamblewithahighprobabilityo
BimaatyoBimfalargerloss,diminishingsensitivitymakesthesurelossmoreaversive,
and the certainty effect reduces the aversiveness of the gamble. The same two factors
enhancetheattractiveness ofthesure thingandreduce theattractivenessof thegamble
whentheoutcomesarepositive.
The shape of the value function and the decision weights both contribute to the
patternobservedinthetoprowoftable13.Inthebottomrow,however,thetwofactors
operateinoppositedirections:diminishingsensitivitycontinuestofavorriskaversionfor
gainsand riskseeking for losses,but theoverweighting oflow probabilities overcomes
thiseffectandproducestheobservedpatternofgamblingforgainsandcautionforlosses.
Manyunfortunatehumansituationsunfoldinthetoprightcell.Thisiswherepeople
whofaceverybadoptionstakedesperategambles,acceptingahighprobabilityofmaking
thingsworseinexchangeforasmallhopeofavoidingalargeloss.Risktakingofthiskind
oftenturnsmanageablefailuresintodisasters.Thethoughtofacceptingthelargesureloss
istoopainful,andthehopeofcompleterelieftooenticing,tomakethesensibledecision
that it is time to cut one’s losses. This is where businesses that are losing ground to a
superiortechnologywastetheirremainingassetsinfutileattemptstocatchup.Because
defeatissodifficulttoaccept,thelosingsideinwarsoftenfightslongpastthepointat
whichthevictoryoftheothersideiscertain,andonlyamatteroftime.
GamblingintheShadowoftheLaw
The legal scholar Chris Guthrie has offered a compelling application of the fourfold
patterntotwosituationsinwhichtheplaintiffandthedefendantinacivilsuitconsidera
possiblesettlement.Thesituationsdifferinthestrengthoftheplaintiff’scase.
Asinascenariowesawearlier,youaretheplaintiffinacivilsuitinwhichyouhave
madeaclaimforalargesumindamages.Thetrialisgoingverywellandyourlawyer
cites expert opinion that you have a 95% chance to win outright, but adds the caution,
“Youneverreallyknowtheoutcomeuntilthejurycomesin.”Yourlawyerurgesyouto
acceptasettlementinwhichyoumightgetonly90%ofyourclaim.Youareinthetopleft
cellofthefourfoldpattern,andthequestiononyourmindis,“AmIwillingtotakeevena
smallchanceofgettingnothingatall?Even90%oftheclaimisagreatdealofmoney,
andIcan walkawaywith it now.” Twoemotionsare evoked,bothdriving in thesame
direction: the attraction of a sure (and substantial) gain and the fear of intense
disappointmentand regretif youreject asettlement and lose incourt. Youcan feel the
pressure that typically leads to cautious behavior in this situation. The plaintiff with a
strongcaseislikelytoberiskaverse.
Nowstepintotheshoesofthedefendantinthesamecase.Althoughyouhavenot
completelygivenuphopeofadecisioninyourfavor,yourealizethatthetrialisgoing
poorly.Theplaintiff’slawyershaveproposedasettlementinwhichyouwouldhavetopay
90%oftheiroriginalclaim,anditiscleartheywillnotacceptless.Willyousettle,orwill
youpursuethecase?Becauseyoufaceahighprobabilityofaloss,yoursituationbelongs
inthetoprightcell.Thetemptationtofightonisstrong:thesettlementthattheplaintiff
hasofferedisalmostaspainfulastheworstoutcomeyouface,andthereisstillhopeof
prevailingincourt.Hereagain,twoemotionsareinvolved:thesurelossisrepugnantand
thepossibilityofwinningincourtishighlyattractive.Adefendantwithaweakcaseis
likelyto berisk seeking,Bima aing,Bim preparedto gamblerather thanaccept avery
unfavorablesettlement.Intheface-offbetweenarisk-averseplaintiffandarisk-seeking
defendant,thedefendantholdsthestrongerhand.Thesuperiorbargainingpositionofthe
defendantshouldbereflectedinnegotiatedsettlements,withtheplaintiffsettlingforless
than the statistically expected outcome of the trial. This prediction from the fourfold
patternwasconfirmedbyexperimentsconductedwithlawstudentsandpracticingjudges,
andalsobyanalysesofactualnegotiationsintheshadowofciviltrials.
Nowconsider“frivolouslitigation,”whenaplaintiffwithaflimsycasefilesalarge
claimthatismostlikelytofailincourt.Bothsidesareawareoftheprobabilities,andboth
know that in a negotiated settlement the plaintiff will get only a small fraction of the
amount of the claim. The negotiation is conducted in the bottom row of the fourfold
pattern. The plaintiff is in the left-hand cell, with a small chance to win a very large
amount;thefrivolousclaimisalotteryticketforalargeprize.Overweightingthesmall
chanceofsuccessisnaturalinthissituation,leadingtheplaintifftobeboldandaggressive
inthenegotiation.Forthedefendant,thesuitisanuisancewithasmallriskofaverybad
outcome.Overweightingthesmallchanceofalargelossfavorsriskaversion,andsettling
foramodestamountisequivalenttopurchasinginsuranceagainsttheunlikelyeventofa
badverdict.Theshoeisnowontheotherfoot:theplaintiffiswillingtogambleandthe
defendant wants to be safe. Plaintiffs with frivolous claims are likely to obtain a more
generoussettlementthanthestatisticsofthesituationjustify.
Thedecisionsdescribedbythefourfoldpatternarenotobviouslyunreasonable.You
canempathizeineachcasewiththefeelingsoftheplaintiffandthedefendantthatlead
them to adopt a combative or an accommodating posture. In the long run, however,
deviationsfromexpectedvaluearelikelytobecostly.Consideralargeorganization,the
CityofNewYork,andsupposeitfaces200“frivolous”suitseachyear,eachwitha5%
chancetocostthecity$1million.Supposefurtherthatineachcasethecitycouldsettle
thelawsuitforapaymentof$100,000.Thecityconsiderstwoalternativepoliciesthatit
willapplytoallsuchcases:settleorgototrial.(Forsimplicity,Iignorelegalcosts.)
Ifthecitylitigatesall200cases,itwilllose10,foratotallossof$10million.
Ifthecitysettleseverycasefor$100,000,itstotallosswillbe$20million.
When you take the long view of many similar decisions, you can see that paying a
premiumtoavoidasmallriskofalargelossiscostly.Asimilaranalysisappliestoeachof
thecellsofthefourfoldpattern:systematicdeviationsfromexpectedvaluearecostlyin
the long run—and this rule applies to both risk aversion and risk seeking. Consistent
overweighting of improbable outcomes—a feature of intuitive decision making—
eventuallyleadstoinferioroutcomes.
SpeakingOfTheFourfoldPattern
“Heistemptedtosettlethisfrivolousclaimtoavoidafreakloss,howeverunlikely.
That’soverweightingofsmallprobabilities.Sinceheislikelytofacemanysimilar
problems,hewouldbebetteroffnotyielding.”
“WeneverletourvacationshangBimaaangBimonalast-minutedeal.We’rewilling
topayalotforcertainty.”
“Theywillnotcuttheirlossessolongasthereisachanceofbreakingeven.Thisis
risk-seekinginthelosses.”
“Theyknowtheriskofagasexplosionisminuscule,buttheywantitmitigated.It’sa
possibilityeffect,andtheywantpeaceofmind.”
P
RareEvents
I visited Israel several times during a period in which suicide bombings in buses were
relativelycommon—thoughofcoursequiterareinabsoluteterms.Therewerealtogether
23bombingsbetweenDecember2001andSeptember2004,whichhadcausedatotalof
236fatalities.ThenumberofdailybusridersinIsraelwasapproximately1.3millionat
thattime.Foranytraveler,therisksweretiny,butthatwasnothowthepublicfeltaboutit.
Peopleavoidedbusesasmuchastheycould,andmanytravelersspenttheirtimeonthe
bus anxiously scanning their neighbors for packages or bulky clothes that might hide a
bomb.
Ididnothavemuchoccasiontotravelonbuses,asIwasdrivingarentedcar,butI
waschagrinedtodiscoverthatmybehaviorwasalsoaffected.IfoundthatIdidnotliketo
stopnexttoabusataredlight,andIdroveawaymorequicklythanusualwhenthelight
changed.Iwasashamedofmyself,becauseofcourseIknewbetter.Iknewthattherisk
wastrulynegligible,andthatanyeffectatallonmyactionswouldassignaninordinately
high“decisionweight”toaminusculeprobability.Infact,Iwasmorelikelytobeinjured
in a driving accident than by stopping near a bus. But my avoidance of buses was not
motivatedbya rational concernforsurvival.What drovemewas the experienceofthe
moment: being next to a bus made me think of bombs, and these thoughts were
unpleasant.IwasavoidingbusesbecauseIwantedtothinkofsomethingelse.
Myexperienceillustrateshowterrorismworksandwhyitissoeffective:itinducesan
availabilitycascade.Anextremelyvividimageofdeathanddamage,constantlyreinforced
bymediaattentionandfrequentconversations,becomeshighlyaccessible,especiallyifit
isassociatedwithaspecificsituationsuchasthesightofabus.Theemotionalarousalis
associative,automatic,anduncontrolled,anditproducesanimpulseforprotectiveaction.
System2may“know”thattheprobabilityislow,butthisknowledgedoesnoteliminate
theself-generateddiscomfortandthewishtoavoidit.System1cannotbeturnedoff.The
emotionisnotonlydisproportionatetotheprobability,itisalsoinsensitivetotheexact
level of probability. Suppose that two cities have been warned about the presence of
suicide bombers. Residents of one city are told that two bombers are ready to strike.
Residentsofanothercityaretoldofasinglebomber.Theirriskislowerbyhalf,butdo
theyfeelmuchsafer?
ManystoresinNewYorkCityselllotterytickets,andbusinessisgood.Thepsychology
ofhigh-prizelotteriesissimilartothepsychologyofterrorism.Thethrillingpossibilityof
winningthebigprizeissharedbythecommunityandreCmuninforcedbyconversations
atworkandathome.Buyingaticketisimmediatelyrewardedbypleasantfantasies,just
asavoidingabuswasimmediatelyrewardedbyrelieffromfear.Inbothcases,theactual
probability is inconsequential; only possibility matters. The original formulation of
prospecttheoryincludedtheargumentthat“highlyunlikelyeventsareeitherignoredor
overweighted,” but it did not specify the conditions under which one or the other will
occur,nordiditproposeapsychologicalinterpretationofit.Mycurrentviewofdecision
weights has been strongly influenced by recent research on the role of emotions and
vividnessindecisionmaking.OverweightingofunlikelyoutcomesisrootedinSystem1
featuresthat arefamiliar bynow.Emotion and vividness influence fluency, availability,
andjudgmentsofprobability—andthusaccountforourexcessiveresponsetothefewrare
eventsthatwedonotignore.
OverestimationandOverweighting
WhatisyourjudgmentoftheprobabilitythatthenextpresidentoftheUnitedStates
willbeathird-partycandidate?
Howmuchwillyoupayforabetinwhichyoureceive$1,000ifthenextpresidentof
theUnitedStatesisathird-partycandidate,andnomoneyotherwise?
The two questions are different but obviously related. The first asks you to assess the
probabilityofanunlikelyevent.Thesecondinvitesyoutoputadecisionweightonthe
sameevent,byplacingabetonit.
Howdopeoplemakethe judgments andhowdotheyassign decision weights?We
startfromtwosimpleanswers,thenqualifythem.Herearetheoversimplifiedanswers:
Peopleoverestimatetheprobabilitiesofunlikelyevents.
Peopleoverweightunlikelyeventsintheirdecisions.
Although overestimation and overweighting are distinct phenomena, the same
psychologicalmechanismsareinvolvedinboth:focusedattention,confirmationbias,and
cognitiveease.
Specific descriptions trigger the associative machinery of System 1. When you
thought about the unlikely victory of a third-party candidate, your associative system
worked in its usual confirmatory mode, selectively retrieving evidence, instances, and
images that would make the statement true. The process was biased, but it was not an
exerciseinfantasy.Youlookedforaplausiblescenariothatconformstotheconstraintsof
reality;youdidnotsimplyimaginetheFairyoftheWestinstallingathird-partypresident.
Yourjudgmentofprobabilitywasultimatelydeterminedbythecognitiveease,orfluency,
withwhichaplausiblescenariocametomind.
Youdonotalwaysfocusontheeventyouareaskedtoestimate.Ifthetargeteventis
verylikely,youfocusonitsalternative.Considerthisexample:
Whatistheprobabilitythatababyborninyourlocalhospitalwillbereleasedwithin
threedays?
You were asked to estimate the probability of the baby going home, but you almost
certainly focused on the events that might cause a baby not to be released within the
normalperiod.OurmindhasausefulcapabilitytoBmunqtoBmufocusspontaneouslyon
whateverisodd,different,orunusual.Youquicklyrealizedthatitisnormalforbabiesin
theUnitedStates(notallcountrieshavethesamestandards)tobereleasedwithintwoor
threedaysofbirth,soyourattentionturnedtotheabnormalalternative.Theunlikelyevent
became focal. The availability heuristic is likely to be evoked: your judgment was
probablydeterminedbythenumberofscenariosofmedicalproblemsyouproducedand
bytheeasewithwhichtheycametomind.Becauseyouwereinconfirmatorymode,there
isagoodchancethatyourestimateofthefrequencyofproblemswastoohigh.
Theprobabilityofarareeventismostlikelytobeoverestimatedwhenthealternative
isnotfullyspecified.MyfavoriteexamplecomesfromastudythatthepsychologistCraig
FoxconductedwhilehewasAmos’sstudent.Foxrecruitedfansofprofessionalbasketball
andelicitedseveraljudgmentsanddecisionsconcerningthewinneroftheNBAplayoffs.
Inparticular,heaskedthemtoestimatetheprobabilitythateachoftheeightparticipating
teamswouldwintheplayoff;thevictoryofeachteaminturnwasthefocalevent.
You can surely guess what happened, but the magnitude of the effect that Fox
observedmaysurpriseyou.Imagineafanwhohasbeenaskedtoestimatethechancesthat
the Chicago Bulls will win the tournament. The focal event is well defined, but its
alternative—one of the other seven teams winning—is diffuse and less evocative. The
fan’smemoryandimagination,operatinginconfirmatorymode,aretryingtoconstructa
victory for the Bulls. When the same person is next asked to assess the chances of the
Lakers, the same selective activation will work in favor of that team. The eight best
professionalbasketballteamsintheUnitedStatesareallverygood,anditispossibleto
imagineevenarelativelyweakteamamongthememergingaschampion.Theresult:the
probabilityjudgmentsgeneratedsuccessivelyfortheeightteamsaddedupto240%!This
patternisabsurd,ofcourse,becausethesumofthechancesoftheeighteventsmustadd
up to 100%. The absurdity disappeared when the same judges were asked whether the
winner would be from the Eastern or the Western conference. The focal event and its
alternativewereequallyspecificinthatquestionandthejudgmentsoftheirprobabilities
addedupto100%.
To assess decision weights, Fox also invited the basketball fans to bet on the
tournamentresult.Theyassignedacashequivalenttoeachbet(acashamountthatwas
justasattractiveasplayingthebet).Winningthebetwouldearnapayoffof$160.The
sum of the cash equivalents for the eight individual teams was $287. An average
participantwhotookalleightbetswouldbeguaranteedalossof$127!Theparticipants
surelyknewthattherewereeightteamsinthetournamentandthattheaveragepayofffor
bettingonallofthemcouldnotexceed$160,buttheyoverweightednonetheless.Thefans
not only overestimated the probability of the events they focused on—they were also
muchtoowillingtobetonthem.
These findings shed new light on the planning fallacy and other manifestations of
optimism.Thesuccessfulexecutionofaplanisspecificand easy to imaginewhenone
triestoforecasttheoutcomeofaproject.Incontrast,thealternativeoffailureisdiffuse,
because there are innumerable ways for things to go wrong. Entrepreneurs and the
investorswhoevaluatetheirprospectsarepronebothtooverestimatetheirchancesandto
overweighttheirestimates.
VividOutcomes
As we have seen, prospect theory differs from utility theory in the rel Bmun q rel
Bmuationship it suggests between probability and decision weight. In utility theory,
decision weights and probabilities are the same. The decision weight of a sure thing is
100,andtheweightthatcorrespondstoa90%chanceisexactly90,whichis9timesmore
thanthedecisionweight for a 10% chance.In prospecttheory,variationsofprobability
havelesseffectondecisionweights.AnexperimentthatImentionedearlierfoundthatthe
decisionweightfora90%chancewas71.2andthedecisionweightfora10%chancewas
18.6.Theratiooftheprobabilitieswas9.0,buttheratioofthedecisionweightswasonly
3.83, indicating insufficient sensitivity to probability in that range. In both theories, the
decision weights depend only on probability, not on the outcome. Both theories predict
thatthedecisionweightfora90%chanceisthesameforwinning$100,receivingadozen
roses,orgettinganelectricshock.Thistheoreticalpredictionturnsouttobewrong.
PsychologistsattheUniversityofChicagopublishedanarticlewiththeattractivetitle
“Money,Kisses,andElectricShocks:OntheAffectivePsychologyofRisk.”Theirfinding
was that the valuation of gambles was much less sensitive to probability when the
(fictitious)outcomeswereemotional(“meetingandkissingyourfavoritemoviestar”or
“gettingapainful,butnotdangerous,electricshock”)thanwhentheoutcomesweregains
or losses of cash. This was not an isolated finding. Other researchers had found, using
physiologicalmeasuressuchasheartrate,thatthefearofanimpendingelectricshockwas
essentiallyuncorrelatedwiththeprobabilityofreceivingtheshock.Themerepossibility
ofashocktriggeredthefull-blownfearresponse.TheChicagoteamproposedthat“affect-
laden imagery” overwhelmed the response to probability. Ten years later, a team of
psychologistsatPrincetonchallengedthatconclusion.
The Princeton team argued that the low sensitivity to probability that had been
observed for emotional outcomes is normal. Gambles on money are the exception. The
sensitivitytoprobabilityisrelativelyhighforthesegambles,becausetheyhaveadefinite
expectedvalue.
Whatamountofcashisasattractiveaseachofthesegambles?
A.84%chancetowin$59
B.84%chancetoreceiveonedozenredrosesinaglassvase
Whatdoyounotice?ThesalientdifferenceisthatquestionAismucheasierthanquestion
B. You did not stop to compute the expected value of the bet, but you probably knew
quickly that it is not far from $50 (in fact it is $49.56), and the vague estimate was
sufficienttoprovideahelpfulanchorasyousearchedforanequallyattractivecashgift.
No such anchor is available for question B, which is therefore much harder to answer.
Respondentsalsoassessedthecashequivalentofgambleswitha21%chancetowinthe
two outcomes. As expected, the difference between the high-probability and low-
probabilitygambleswasmuchmorepronouncedforthemoneythanfortheroses.
Tobolstertheir argument thatinsensitivitytoprobabilityis notcausedbyemotion,
thePrincetonteamcomparedwillingnesstopaytoavoidgambles:
21%chance(or84%chance)tospendaweekendpaintingsomeone’sthree-bedroom
apartment
21%chance(or84%chance)tocleanthreestallsinadormitorybathBmunqbath
Bmuroomafteraweekendofuse
The second outcome is surely much more emotional than the first, but the decision
weightsforthetwooutcomesdidnotdiffer.Evidently,theintensityofemotionisnotthe
answer.
Another experiment yielded a surprising result. The participants received explicit
priceinformationalongwiththeverbaldescriptionoftheprize.Anexamplecouldbe:
84%chancetowin:Adozenredrosesinaglassvase.Value$59.
21%chancetowin:Adozenredrosesinaglassvase.Value$59.
Itiseasytoassesstheexpectedmonetaryvalueofthesegambles,butaddingaspecific
monetary value did not alter the results: evaluations remained insensitive to probability
eveninthatcondition.Peoplewhothoughtofthegiftasachancetogetrosesdidnotuse
priceinformationasananchorinevaluatingthegamble.Asscientistssometimessay,this
isasurprisingfindingthatistryingtotellussomething.Whatstoryisittryingtotellus?
Thestory,Ibelieve,isthatarichandvividrepresentationoftheoutcome,whetheror
not it is emotional, reduces the role of probability in the evaluation of an uncertain
prospect. This hypothesis suggests a prediction, in which I have reasonably high
confidence: adding irrelevant but vivid details to a monetary outcome also disrupts
calculation.Compareyourcashequivalentsforthefollowingoutcomes:
21%(or84%)chancetoreceive$59nextMonday
21%(or84%)chancetoreceivealargebluecardboardenvelopecontaining$59next
Mondaymorning
Thenewhypothesisisthattherewillbelesssensitivitytoprobabilityinthesecondcase,
becausetheblueenvelopeevokesaricherandmorefluentrepresentationthantheabstract
notionofasumofmoney.Youconstructedtheeventinyourmind,andthevividimageof
the outcome exists there even if you know that its probability is low. Cognitive ease
contributestothecertaintyeffectaswell:whenyouholdavividimageofanevent,the
possibility of its not occurring is also represented vividly, and overweighted. The
combination of an enhanced possibility effect with an enhanced certainty effect leaves
littleroomfordecisionweightstochangebetweenchancesof21%and84%.
VividProbabilities
Theideathatfluency,vividness,andtheeaseofimaginingcontributetodecisionweights
gainssupportfrommanyotherobservations.Participantsinawell-knownexperimentare
given a choice of drawing a marble from one of two urns, in which red marbles win a
prize:
UrnAcontains10marbles,ofwhich1isred.
UrnBcontains100marbles,ofwhich8arered.
Whichurnwouldyouchoose?Thechancesofwinningare10%inurnAand8%inurnB,
so making the right choice should be easy, but it is not: about 30%–40% of students
choosetheurnBmunqurnBmuwiththelargernumberofwinningmarbles,ratherthan
the urn that provides a better chance of winning. Seymour Epstein has argued that the
resultsillustratethesuperficialprocessingcharacteristicofSystem1(whichhecallsthe
experientialsystem).
As you might expect, the remarkably foolish choices that people make in this
situationhaveattractedtheattentionofmanyresearchers.Thebiashasbeengivenseveral
names;followingPaulSlovicIwillcallitdenominatorneglect.Ifyourattentionisdrawn
to the winning marbles, you do not assess the number of nonwinning marbles with the
samecare.Vividimagerycontributestodenominatorneglect,atleastasIexperienceit.
WhenIthinkofthesmallurn,Iseeasingleredmarbleonavaguelydefinedbackground
ofwhitemarbles.WhenIthinkofthelargerurn,Iseeeightwinningredmarblesonan
indistinct background of white marbles, which creates a more hopeful feeling. The
distinctivevividnessofthewinningmarblesincreasesthedecisionweightofthatevent,
enhancingthepossibilityeffect.Ofcourse,thesamewillbetrueofthecertaintyeffect.IfI
havea90%chanceofwinningaprize,theeventofnotwinningwillbemoresalientif10
of100marblesare“losers”thanif1of10marblesyieldsthesameoutcome.
Theideaofdenominatorneglecthelpsexplainwhydifferentwaysofcommunicating
risksvarysomuchintheireffects.Youreadthat“avaccinethatprotectschildrenfroma
fataldiseasecarriesa0.001%riskofpermanentdisability.”Theriskappearssmall.Now
consideranotherdescriptionofthesamerisk:“Oneof100,000vaccinatedchildrenwillbe
permanentlydisabled.”Thesecondstatementdoessomethingtoyourmindthatthefirst
doesnot:itcallsupthe image ofanindividualchildwhoispermanentlydisabled by a
vaccine; the 999,999 safely vaccinated children have faded into the background. As
predictedbydenominatorneglect,low-probabilityeventsaremuchmoreheavilyweighted
when described in terms of relative frequencies (how many) than when stated in more
abstract terms of “chances,” “risk,” or “probability” (how likely). As we have seen,
System1ismuchbetteratdealingwithindividualsthancategories.
Theeffectofthefrequencyformatislarge.Inonestudy,peoplewhosawinformation
about“adiseasethatkills1,286peopleoutofevery10,000”judgeditasmoredangerous
thanpeoplewhoweretoldabout“adiseasethatkills24.14%ofthepopulation.”Thefirst
diseaseappearsmorethreateningthanthesecond,althoughtheformerriskisonlyhalfas
large as the latter! In an even more direct demonstration of denominator neglect, “a
diseasethatkills1,286 peopleoutof every 10,000”wasjudged moredangerousthana
diseasethat“kills24.4outof100.”Theeffectwouldsurelybereducedoreliminatedif
participants were asked for a direct comparison of the two formulations, a task that
explicitlycallsforSystem2.Life,however,isusuallyabetween-subjectsexperiment,in
which you see only one formulation at a time. It would take an exceptionally active
System2togeneratealternativeformulationsoftheoneyouseeandtodiscoverthatthey
evokeadifferentresponse.
Experiencedforensicpsychologistsandpsychiatristsarenotimmunetotheeffectsof
the format in which risks are expressed. In one experiment, professionals evaluated
whetheritwassafetodischargefromthepsychiatrichospitalapatient,Mr.Jones,witha
historyofviolence.Theinformationtheyreceivedincludedanexpert’sassessmentofthe
risk.Thesamestatisticsweredescribedintwoways:
PatientssimilartoMr.Jonesareestimatedtohavea10%probabilityofcommitting
anactofviolenceagainstothersduringthefirstseveralmonthsafterdischarge.
Of every 100 patients similar to Mr. Jones, 10 are estimated to commit an act of
violenceagainstothersduringthefirstseveralmonthsafterdischarge.
Theprofessionalswhosawthefrequencyformatwerealmosttwiceaslikelytodenythe
discharge(41%,comparedto21%intheprobabilityformat).Themorevividdescription
producesahigherdecisionweightforthesameprobability.
Thepowerofformatcreatesopportunitiesformanipulation,whichpeoplewithanaxe
to grind know how to exploit. Slovic and his colleagues cite an article that states that
“approximately1,000homicidesayeararecommittednationwidebyseriouslymentally
illindividualswhoarenottakingtheirmedication.”Anotherwayofexpressingthesame
fact is that “1,000 out of 273,000,000 Americans will die in this manner each year.”
Another is that “the annual likelihood of being killed by such an individual is
approximately0.00036%.”Stillanother:“1,000Americanswilldieinthismannereach
year,orlessthanone-thirtieththenumberwhowilldieofsuicideandaboutone-fourththe
numberwhowilldieoflaryngealcancer.”Slovicpointsoutthat“theseadvocatesarequite
openabouttheirmotivation:they want tofrighten thegeneral publicaboutviolence by
peoplewithmentaldisorder,inthehopethatthisfearwilltranslateintoincreasedfunding
formentalhealthservices.”
AgoodattorneywhowishestocastdoubtonDNAevidencewillnottellthejurythat
“thechanceofafalsematchis0.1%.”Thestatementthat“afalsematchoccursin1of
1,000capitalcases”isfarmorelikelytopassthethresholdofreasonabledoubt.Thejurors
hearingthosewordsareinvitedtogeneratetheimageofthemanwhositsbeforethemin
thecourtroombeingwronglyconvictedbecauseofflawedDNAevidence.Theprosecutor,
of course, will favor the more abstract frame—hoping to fill the jurors’ minds with
decimalpoints.
DecisionsfromGlobalImpressions
Theevidencesuggeststhehypothesisthatfocalattentionandsaliencecontributetoboth
the overestimation of unlikely events and the overweighting of unlikely outcomes.
Salienceisenhancedbymerementionofanevent,byitsvividness,andbytheformatin
whichprobabilityisdescribed.Thereareexceptions,ofcourse,inwhichfocusingonan
eventdoesnot raiseitsprobability:casesinwhichanerroneoustheorymakesanevent
appearimpossibleevenwhenyouthinkaboutit,orcasesinwhichaninabilitytoimagine
howanoutcomemightcomeaboutleavesyouconvincedthatitwillnothappen.Thebias
towardoverestimationandoverweightingofsalienteventsisnotanabsoluterule,butitis
largeandrobust.
There has been much interest in recent years in studies of choice from experience,
which follow different rules from the choices from description that are analyzed in
prospecttheory.Participantsinatypicalexperimentfacetwobuttons.Whenpressed,each
buttonproduceseitheramonetaryrewardornothing,andtheoutcomeisdrawnrandomly
according to the specifications of a prospect (for example, “5% to win $12” or “95%
chance to win $1”). The process is truly random, s Bmun qm, s Bmuo there is no
guarantee that the sample a participant sees exactly represents the statistical setup. The
expectedvaluesassociatedwiththetwobuttonsareapproximatelyequal,butoneisriskier
(morevariable)thantheother.(Forexample,onebuttonmayproduce$10on5%ofthe
trialsandtheother$1on50%ofthetrials).Choicefromexperienceisimplementedby
exposing the participant to many trials in which she can observe the consequences of
pressingonebuttonoranother.Onthecriticaltrial,shechoosesoneofthetwobuttons,
andsheearnstheoutcomeonthattrial.Choicefromdescriptionisrealizedbyshowingthe
subjecttheverbaldescriptionoftheriskyprospectassociatedwitheachbutton(suchas
“5%towin$12”)andaskinghertochooseone.Asexpectedfromprospecttheory,choice
from description yields a possibility effect—rare outcomes are overweighted relative to
their probability. In sharp contrast, overweighting is never observed in choice from
experience,andunderweightingiscommon.
The experimental situation of choice by experience is intended to represent many
situations in which we are exposed to variable outcomes from the same source. A
restaurantthatisusuallygoodmayoccasionallyserveabrilliantoranawfulmeal.Your
friendisusuallygoodcompany,buthesometimesturnsmoodyandaggressive.California
ispronetoearthquakes,buttheyhappenrarely.Theresultsofmanyexperimentssuggest
that rare events are not overweighted when we make decisions such as choosing a
restaurantortyingdowntheboilertoreduceearthquakedamage.
The interpretation of choice from experience is not yet settled, but there is general
agreementononemajorcauseofunderweightingofrareevents,bothinexperimentsand
inthe realworld:many participantsnever experiencetherare event!Most Californians
have never experienced a major earthquake, and in 2007 no banker had personally
experiencedadevastatingfinancialcrisis.RalphHertwigandIdoErevnotethat“chances
ofrareevents(suchastheburstofhousingbubbles)receivelessimpactthantheydeserve
according to their objective probabilities.” They point to the public’s tepid response to
long-termenvironmentalthreatsasanexample.
These examples of neglect are both important and easily explained, but
underweighting also occurs when people have actually experienced the rare event.
Suppose you have a complicated question that two colleagues on your floor could
probablyanswer.Youhaveknownthembothforyearsandhavehadmanyoccasionsto
observe and experience their character. Adele is fairly consistent and generally helpful,
though not exceptional on that dimension. Brian is not quite as friendly and helpful as
Adelemostofthetime,butonsomeoccasionshehasbeenextremelygenerouswithhis
timeandadvice.Whomwillyouapproach?
Considertwopossibleviewsofthisdecision:
Itisachoicebetweentwogambles.Adeleisclosertoasurething;theprospectof
Brianismorelikelytoyieldaslightlyinferioroutcome,withalowprobabilityofa
verygoodone.Therareeventwillbeoverweightedbyapossibilityeffect,favoring
Brian.
ItisachoicebetweenyourglobalimpressionsofAdeleandBrian.Thegoodandthe
bad experiences you have had are pooled in your representation of their normal
behavior.Unlesstherareeventissoextremethatitcomestomindseparately(Brian
onceverballyabusedacolleaguewhoaskedforhishelp),thenormwillbebiased
towardtypicalandrecentinstances,favoringAdele.
In a two-system mind, the second interpretation a Bmun qon a Bmuppears far more
plausible.System1generatesglobalrepresentationsofAdeleandBrian,whichincludean
emotionalattitudeandatendencytoapproachoravoid.Nothingbeyondacomparisonof
thesetendenciesisneededtodeterminethedooronwhichyouwillknock.Unlesstherare
eventcomestoyourmindexplicitly,itwillnotbeoverweighted.Applyingthesameidea
to the experiments on choice from experience is straightforward. As they are observed
generating outcomes over time, the two buttons develop integrated “personalities” to
whichemotionalresponsesareattached.
The conditions under which rare events are ignored or overweighted are better
understoodnowthantheywerewhenprospecttheorywasformulated.Theprobabilityofa
rareeventwill(often,notalways)beoverestimated,becauseoftheconfirmatorybiasof
memory.Thinkingaboutthatevent,youtrytomakeittrueinyourmind.Arareeventwill
be overweighted if it specifically attracts attention. Separate attention is effectively
guaranteedwhenprospectsaredescribedexplicitly(“99%chancetowin$1,000,and1%
chance to win nothing”). Obsessive concerns (the bus in Jerusalem), vivid images (the
roses), concrete representations (1 of 1,000), and explicit reminders (as in choice from
description)all contributeto overweighting.And whenthere isno overweighting,there
willbeneglect.Whenitcomestorareprobabilities,ourmindisnotdesignedtogetthings
quite right. For the residents of a planet that may be exposed to events no one has yet
experienced,thisisnotgoodnews.
SpeakingofRareEvents
“TsunamisareveryrareeveninJapan,buttheimageissovividandcompellingthat
touristsareboundtooverestimatetheirprobability.”
“It’s the familiar disaster cycle. Begin by exaggeration and overweighting, then
neglectsetsin.”
“We shouldn’t focus on a single scenario, or we will overestimate its probability.
Let’ssetupspecificalternativesandmaketheprobabilitiesaddupto100%.”
“Theywantpeopletobeworriedbytherisk.That’swhytheydescribeitas1death
per1,000.They’recountingondenominatorneglect.”
P
RiskPolicies
Imagine that you face the following pair of concurrent decisions. First examine both
decisions,thenmakeyourchoices.
Decision(i):Choosebetween
A.suregainof$240
B.25%chancetogain$1,000and75%chancetogainnothing
Decision(ii):Choosebetween
C.surelossof$750
D.75%chancetolose$1,000and25%chancetolosenothing
Thispairofchoiceproblemshasanimportantplaceinthehistoryofprospecttheory,and
it has new things to tell us about rationality. As you skimmed the two problems, your
initialreactiontothesurethings(AandC)wasattractiontothefirstandaversiontothe
second.Theemotionalevaluationof“suregain”and“sureloss”isanautomaticreaction
ofSystem1,whichcertainlyoccursbeforethemoreeffortful(andoptional)computation
of the expected values of the two gambles (respectively, a gain of $250 and a loss of
$750). Most people’s choices correspond to the predilections of System 1, and large
majoritiespreferAtoBandDtoC.Asinmanyotherchoicesthatinvolvemoderateor
highprobabilities,peopletendtoberiskaverseinthedomainofgainsandriskseekingin
the domain of losses. In the original experiment that Amos and I carried out, 73% of
respondents chose A in decision i and D in decision ii and only 3% favored the
combinationofBandC.
You were asked to examine both options before making your first choice, and you
probablydidso.Butonethingyousurelydidnotdo:youdidnotcomputethepossible
results of the four combinations of choices (A and C, A and D, B and C, B and D) to
determine which combination you like best. Your separate preferences for the two
problemswereintuitivelycompellingandtherewasnoreasontoexpectthattheycould
leadtotrouble.Furthermore,combiningthetwodecisionproblemsisalaboriousexercise
thatyouwouldneedpaperandpenciltocomplete.Youdidnotdoit.Nowconsiderthe
followingchoiceproblem:
AD.25%chancetowin$240and75%chancetolose$760
BC.25%chancetowin$250and75%chancetolose$750
Thischoiceiseasy!OptionBCactuallydominatesoptionAD(thetechnicaltermforone
optionbeingunequivocallybetterthananother).Youalreadyknowwhatcomesnext.The
dominantoptioninADisthecombinationofthetworejectedoptionsinthefirstpairof
decisionproblems,theonethatonly3%ofrespondentsfavoredinouroriginalstudy.The
inferioroptionBCwaspreferredby73%ofrespondents.
BroadorNarrow?
Thissetofchoiceshasalottotellusaboutthelimitsofhumanrationality.Foronething,
ithelps us see the logicalconsistency of Human preferences forwhat it is—a hopeless
mirage.Haveanotherlookatthelastproblem,theeasyone.Wouldyouhaveimaginedthe
possibilityofdecomposingthisobviouschoiceproblemintoapairofproblemsthatwould
leadalargemajorityofpeopletochooseaninferioroption?Thisisgenerallytrue:every
simplechoiceformulatedintermsofgainsandlossescanbedeconstructedininnumerable
waysintoacombinationofchoices,yieldingpreferencesthatarelikelytobeinconsistent.
Theexamplealsoshowsthatitiscostlytoberiskaverseforgainsandriskseeking
forlosses.Theseattitudesmakeyouwillingtopayapremiumtoobtainasuregainrather
thanfaceagamble,andalsowillingtopayapremium(inexpectedvalue)toavoidasure
loss. Both payments come out of the same pocket, and when you face both kinds of
problemsatonce,thediscrepantattitudesareunlikelytobeoptimal.
ThereweretwBghthecomeoowaysofconstruingdecisionsiandii:
narrowframing:asequenceoftwosimpledecisions,consideredseparately
broadframing:asinglecomprehensivedecision,withfouroptions
Broadframingwasobviouslysuperiorinthiscase.Indeed,itwillbesuperior(oratleast
not inferior) in every case in which several decisions are to be contemplated together.
Imaginealongerlistof5simple(binary)decisionstobeconsideredsimultaneously.The
broad(comprehensive)frameconsistsofasinglechoicewith32options.Narrowframing
willyieldasequenceof5simplechoices.Thesequenceof5choiceswillbeoneofthe32
optionsof the broad frame. Will it be the best?Perhaps, butnot very likely. Arational
agentwillofcourseengageinbroadframing,butHumansarebynaturenarrowframers.
The ideal of logical consistency, as this example shows, is not achievable by our
limitedmind.BecausewearesusceptibletoWYSIATIandaversetomentaleffort,we
tend to make decisions as problems arise, even when we are specifically instructed to
considerthemjointly.Wehaveneithertheinclinationnorthementalresourcestoenforce
consistencyonourpreferences,andourpreferencesarenotmagicallysettobecoherent,
astheyareintherational-agentmodel.
Samuelson’sProblem
The great Paul Samuelson—a giant among the economists of the twentieth century—
famouslyaskedafriendwhetherhewouldacceptagambleonthetossofacoininwhich
hecouldlose$100orwin$200.Hisfriendresponded,“Iwon’tbetbecauseIwouldfeel
the$100lossmorethanthe$200gain.ButI’lltakeyouonifyoupromisetoletmemake
100 such bets.” Unless you are a decision theorist, you probably share the intuition of
Samuelson’sfriend,thatplayingaveryfavorablebutriskygamblemultipletimesreduces
the subjective risk. Samuelson found his friend’s answer interesting and went on to
analyzeit.Heprovedthatundersomeveryspecificconditions,autilitymaximizerwho
rejectsasinglegambleshouldalsorejecttheofferofmany.
Remarkably, Samuelson did not seem to mind the fact that his proof, which is of
coursevalid,ledtoaconclusionthatviolatescommonsense,ifnotrationality:theofferof
ahundredgamblesissoattractivethatnosanepersonwouldrejectit.MatthewRabinand
Richard Thaler pointed out that “the aggregated gamble of one hundred 50–50 lose
$100/gain $200 bets has an expected return of $5,000, with only a 1/2,300 chance of
losinganymoneyandmerelya1/62,000chanceoflosingmorethan$1,000.”Theirpoint,
ofcourse,isthatifutilitytheorycanbeconsistentwithsuchafoolishpreferenceunder
anycircumstances,thensomethingmustbewrongwithitasamodelofrationalchoice.
SamuelsonhadnotseenRabin’sproofoftheabsurdconsequencesofseverelossaversion
forsmallbets,buthewouldsurelynothavebeensurprisedbyit.Hiswillingnessevento
consider the possibility that it could be rational to reject the package testifies to the
powerfulholdoftherationalmodel.
Let us assume that a very simple value function describes the preferences of
Samuelson’sfriend(callhimSam).ToexpresshisaversiontolossesSamfirstrewritesthe
bet,aftermultiplyingeachlossbyafactorof2.Hethencomputestheexpectedvalueof
therewrittenbet.Herearetheresults,forone,two,orthreetosses.Theyaresufficiently
instructivetodeservesomeBghticiof2
Youcanseeinthedisplaythatthegamblehasanexpectedvalueof50.However,onetoss
isworthnothingtoSambecausehefeelsthatthepainoflosingadollaristwiceasintense
asthepleasureofwinningadollar.Afterrewritingthegambletoreflecthislossaversion,
Samwillfindthatthevalueofthegambleis0.
Nowconsidertwotosses.Thechancesoflosinghavegonedownto25%.Thetwo
extremeoutcomes(lose200orwin400)canceloutinvalue;theyareequallylikely,and
thelossesareweightedtwiceasmuchasthegain.Buttheintermediateoutcome(oneloss,
onegain)ispositive,andsoisthecompoundgambleasawhole.Nowyoucanseethe
cost of narrow framing and the magic of aggregating gambles. Here are two favorable
gambles,whichindividuallyareworthnothingtoSam.Ifheencounterstheofferontwo
separateoccasions,hewillturnitdownbothtimes.However,ifhebundlesthetwooffers
together,theyarejointlyworth$50!
Thingsgetevenbetterwhenthreegamblesarebundled.Theextremeoutcomesstill
cancel out, but they have become less significant. The third toss, although worthless if
evaluatedonitsown,hasadded$62.50tothetotalvalueofthepackage.BythetimeSam
isofferedfive gambles,the expected valueof the offerwill be$250,his probabilityof
losing anything will be 18.75%, and his cash equivalent will be $203.125. The notable
aspect of this story is that Sam never wavers in his aversion to losses. However, the
aggregationoffavorablegamblesrapidlyreducestheprobabilityoflosing,andtheimpact
oflossaversiononhispreferencesdiminishesaccordingly.
NowIhaveasermonreadyforSamifherejectstheofferofasinglehighlyfavorable
gambleplayedonce,andforyouifyousharehisunreasonableaversiontolosses:
Isympathizewithyouraversiontolosinganygamble,butitiscostingyoualotof
money.Pleaseconsiderthisquestion:Areyouonyourdeathbed?Isthisthelastoffer
ofasmallfavorablegamblethatyouwilleverconsider?Ofcourse,youareunlikely
to be offered exactly this gamble again, but you will have many opportunities to
considerattractive gambles withstakes thatare very smallrelative toyour wealth.
You will do yourself a large financial favor if you are able to see each of these
gamblesaspartofabundleofsmallgamblesandrehearsethemantrathatwillget
yousignificantlyclosertoeconomicrationality:youwinafew,youloseafew.The
mainpurposeofthemantraistocontrolyouremotionalresponsewhenyoudolose.
If you can trust it to be effective, you should remind yourself of it when deciding
whetherornottoacceptasmallriskwithpositiveexpectedvalue.Rememberthese
qualificationswhenusingthemantra:
Itworkswhenthegamblesaregenuinelyindependentofeachother;itdoesnotapply
tomultipleinvestmentsinthesameindustry,whichwouldallgobadtogether.
Itworks onlywhenthepossiblelossdoesnotcauseyoutoworryaboutyourtotal
wealth. If you would take the loss as significant bad news about your economic
future,watchit!
Itshouldnotbeappliedtolongshots,wheretheprobabilityofwinningisverysmall
foreachbet.
Ifyouhavetheemotionaldisciplinethatthisrulerequires,Bghtldforeyouwill
neverconsiderasmallgambleinisolationorbelossaverseforasmallgambleuntil
youareactuallyonyourdeathbed—andnoteventhen.
Thisadviceisnotimpossibletofollow.Experiencedtradersinfinancialmarketslive
byiteveryday,shieldingthemselvesfromthepainoflossesbybroadframing.As was
mentionedearlier,wenowknowthatexperimentalsubjectscouldbealmostcuredoftheir
loss aversion (in a particular context) by inducing them to “think like a trader,” just as
experienced baseball card traders are not as susceptible to the endowment effect as
novices are. Students made risky decisions (to accept or reject gambles in which they
couldlose)underdifferentinstructions.Inthenarrow-framingcondition,theyweretoldto
“make each decision as if it were the only one” and to accept their emotions. The
instructionsforbroadframingofadecisionincludedthephrases“imagineyourselfasa
trader,”“youdothisallthetime,”and“treatitasoneofmanymonetarydecisions,which
will sum together to produce a ‘portfolio.’” The experimenters assessed the subjects’
emotionalresponsetogainsandlossesbyphysiologicalmeasures,includingchangesin
theelectrical conductanceof the skin that areused in liedetection. Asexpected, broad
framing blunted the emotional reaction to losses and increased the willingness to take
risks.
The combination of loss aversion and narrow framing is a costly curse. Individual
investorscan avoidthatcurse, achievingthe emotionalbenefitsof broadframing while
alsosavingtimeandagony,byreducingthefrequencywithwhichtheycheckhowwell
theirinvestmentsaredoing.Closelyfollowingdailyfluctuationsisalosingproposition,
becausethepainofthefrequentsmalllossesexceedsthepleasureoftheequallyfrequent
small gains. Once a quarter is enough, and may be more than enough for individual
investors.Inadditiontoimprovingtheemotionalqualityoflife,thedeliberateavoidance
ofexposuretoshort-termoutcomesimprovesthequalityofbothdecisionsandoutcomes.
Thetypicalshort-termreactiontobadnewsisincreasedlossaversion.Investorswhoget
aggregated feedback receive such news much less often and are likely to be less risk
averseandtoendupricher.Youarealsolesspronetouselesschurningofyourportfolioif
youdon’tknowhoweverystockinitisdoingeveryday(oreveryweekorevenevery
month).Acommitmentnottochangeone’spositionforseveralperiods(theequivalentof
“lockingin”aninvestment)improvesfinancialperformance.
RiskPolicies
Decisionmakerswhoarepronetonarrowframingconstructapreferenceeverytimethey
faceariskychoice.Theywoulddobetterbyhavingariskpolicythattheyroutinelyapply
wheneverarelevantproblemarises.Familiarexamplesofriskpoliciesare“alwaystake
the highest possible deductible when purchasing insurance” and “never buy extended
warranties.” A risk policy is a broad frame. In the insurance examples, you expect the
occasionallossoftheentiredeductible,ortheoccasionalfailureofanuninsuredproduct.
Therelevantissueisyourabilitytoreduceoreliminatethepainoftheoccasionallossby
thethoughtthatthepolicythatleftyouexposedtoitwillalmostcertainlybefinancially
advantageousoverthelongrun.
Ariskpolicythataggregatesdecisionsisanalogoustotheoutsideviewofplanning
problemsthatIdiscussedearlier.Theoutsideviewshiftsthefocusfromthespecificsof
thecurrentsituationtoBghtpecicytthestatisticsofoutcomesinsimilarsituations.The
outsideviewisabroadframeforthinkingaboutplans.Ariskpolicyisabroadframethat
embedsaparticularriskychoiceinasetofsimilarchoices.
The outside view and the risk policy are remedies against two distinct biases that
affect many decisions: the exaggerated optimism of the planning fallacy and the
exaggerated caution induced by loss aversion. The two biases oppose each other.
Exaggeratedoptimismprotectsindividualsandorganizationsfromtheparalyzingeffects
oflossaversion;lossaversionprotectsthemfromthefolliesofoverconfidentoptimism.
The upshot is rather comfortable for the decision maker. Optimists believe that the
decisions they make are more prudent than they really are, and loss-averse decision
makerscorrectlyrejectmarginalpropositionsthattheymightotherwiseaccept.Thereis
noguarantee,ofcourse,thatthebiasescanceloutineverysituation.Anorganizationthat
couldeliminatebothexcessive optimism andexcessivelossaversion should doso.The
combinationoftheoutsideviewwithariskpolicyshouldbethegoal.
Richard Thaler tells of a discussion about decision making he had with the top
managersofthe25divisionsofalargecompany.Heaskedthemtoconsiderariskyoption
in which, with equal probabilities, they could lose a large amount of the capital they
controlledorearndoublethatamount.Noneoftheexecutiveswaswillingtotakesucha
dangerousgamble.ThalerthenturnedtotheCEOofthecompany,whowasalsopresent,
andaskedforhisopinion.Withouthesitation,theCEOanswered,“Iwouldlikeallofthem
toaccepttheirrisks.” In the context of thatconversation,itwasnaturalfortheCEOto
adopta broadframethatencompassedall25bets.LikeSamfacing100cointosses,he
couldcountonstatisticalaggregationtomitigatetheoverallrisk.
SpeakingofRiskPolicies
“Tellhertothinklikeatrader!Youwinafew,youloseafew.”
“Idecidedtoevaluatemyportfolioonlyonceaquarter.Iamtoolossaversetomake
sensibledecisionsinthefaceofdailypricefluctuations.”
“Theyneverbuyextendedwarranties.That’stheirriskpolicy.”
“Eachofourexecutivesislossaverseinhisorherdomain.That’sperfectlynatural,
buttheresultisthattheorganizationisnottakingenoughrisk.”
P
KeepingScore
Exceptfortheverypoor,forwhomincomecoincideswithsurvival,themainmotivators
ofmoney-seekingarenotnecessarilyeconomic.Forthebillionairelookingfortheextra
billion,andindeedfortheparticipantinanexperimentaleconomicsprojectlookingforthe
extradollar,moneyisaproxyforpointsonascaleofself-regardandachievement.These
rewardsand punishments, promisesandthreats, areallin ourheads.We carefully keep
score of them. They shape o C Th5ur preferences and motivate our actions, like the
incentivesprovidedinthesocialenvironment.Asaresult,werefusetocutlosseswhen
doingsowouldadmitfailure,wearebiasedagainstactionsthatcouldleadtoregret,and
wedraw anillusory but sharpdistinction betweenomission and commission,not doing
anddoing,becausethe sense ofresponsibilityisgreaterfor onethan fortheother.The
ultimate currency that rewards or punishes is often emotional, a form of mental self-
dealingthatinevitablycreatesconflictsofinterestwhentheindividualactsasanagenton
behalfofanorganization.
MentalAccounts
Richard Thaler has been fascinated for many years by analogies between the world of
accountingandthementalaccountsthatweusetoorganizeandrunourlives,withresults
thataresometimesfoolishandsometimesveryhelpful.Mentalaccountscomeinseveral
varieties. We hold our money in different accounts, which are sometimes physical,
sometimesonlymental.Wehavespendingmoney,generalsavings,earmarkedsavingsfor
our children’s education or for medical emergencies. There is a clear hierarchy in our
willingnesstodrawontheseaccountstocovercurrentneeds.Weuseaccountsforself-
control purposes, as in making a household budget, limiting the daily consumption of
espressos, or increasing the time spent exercising. Often we pay for self-control, for
instance simultaneously putting money in a savings account and maintaining debt on
creditcards. TheEcons ofthe rational-agentmodel donot resortto mentalaccounting:
theyhaveacomprehensiveviewofoutcomesandaredrivenbyexternalincentives.For
Humans,mentalaccountsareaformofnarrowframing;theykeepthingsundercontrol
andmanageablebyafinitemind.
Mentalaccountsareusedextensivelytokeepscore.Recallthatprofessionalgolfers
putt more successfully when working to avoid a bogey than to achieve a birdie. One
conclusionwecandrawisthatthebestgolferscreateaseparateaccountforeachhole;
theydonotonlymaintainasingleaccountfor their overall success. Anironicexample
thatThalerrelatedinanearlyarticleremainsoneofthebestillustrationsofhowmental
accountingaffectsbehavior:
Twoavidsportsfansplantotravel40milestoseeabasketballgame.Oneofthem
paidforhisticket;theotherwasonhiswaytopurchaseaticketwhenhegotonefree
fromafriend.Ablizzardisannouncedforthenightofthegame.Whichofthetwo
ticketholdersismorelikelytobravetheblizzardtoseethegame?
Theanswerisimmediate:weknowthatthefanwhopaidforhisticketismorelikelyto
drive. Mental accounting provides the explanation. We assume that both fans set up an
accountforthegametheyhopedtosee.Missingthegamewillclosetheaccountswitha
negativebalance.Regardlessofhowtheycamebytheirticket,bothwillbedisappointed
—buttheclosingbalanceisdistinctlymorenegativefortheonewhoboughtaticketand
isnowoutofpocketaswellasdeprivedofthegame.Becausestayinghomeisworsefor
thisindividual,heismoremotivatedtoseethegameandthereforemorelikelytomake
theattempttodriveintoablizzard.Thesearetacitcalculationsofemotionalbalance,of
thekindthatSystem1performswithoutdeliberation.Theemotionsthatpeopleattachto
thestateoftheirmentalaccountsarenotacknowledgedinstandardeconomictheory.An
Econwouldrealizethatthetickethasalreadybeenpaidforandcannotbereturned.Its
costis“sunk”andtheEconwouldnotcarewhetherhehadboughtthetickettothegame
orgotitfromafriend(ifEcoBTh5motketnshavefriends).Toimplementthisrational
behavior,System2wouldhavetobeawareofthecounterfactualpossibility:“WouldIstill
driveintothissnowstormifIhadgottentheticketfreefromafriend?”Ittakesanactive
anddisciplinedmindtoraisesuchadifficultquestion.
A related mistake afflicts individual investors when they sell stocks from their
portfolio:
Youneedmoneytocoverthecostsofyourdaughter’sweddingandwillhavetosell
somestock.Yourememberthepriceatwhichyouboughteachstockandcanidentify
itasa“winner,”currentlyworthmorethanyoupaidforit,orasaloser.Amongthe
stocks you own, Blueberry Tiles is a winner; if you sell it today you will have
achievedagainof$5,000.YouholdanequalinvestmentinTiffanyMotors,whichis
currentlyworth$5,000lessthanyoupaidforit.Thevalueofbothstockshasbeen
stableinrecentweeks.Whichareyoumorelikelytosell?
Aplausiblewaytoformulatethechoiceisthis:“IcouldclosetheBlueberryTilesaccount
andscoreasuccessformyrecordasaninvestor.Alternatively,IcouldclosetheTiffany
Motorsaccountandaddafailuretomyrecord.WhichwouldIratherdo?”Iftheproblem
isframedasachoicebetweengivingyourselfpleasureandcausingyourselfpain,youwill
certainlysellBlueberryTilesandenjoyyourinvestmentprowess.Asmightbeexpected,
finance research has documented a massive preference for selling winners rather than
losers—abiasthathasbeengivenanopaquelabel:thedispositioneffect.
Thedispositioneffectisaninstanceofnarrowframing. Theinvestor has setup an
accountforeachsharethatshebought,andshewantstocloseeveryaccountasagain.A
rationalagentwouldhaveacomprehensiveviewoftheportfolioandsellthestockthatis
leastlikelytodowellinthefuture,withoutconsideringwhetheritisawinneroraloser.
Amostoldmeofaconversationwithafinancialadviser,whoaskedhimforacomplete
listofthestocksinhisportfolio,includingthepriceatwhicheachhadbeenpurchased.
When Amos asked mildly, “Isn’t it supposed not to matter?” the adviser looked
astonished.Hehadapparentlyalwaysbelievedthatthestateofthementalaccountwasa
validconsideration.
Amos’s guess about the financial adviser’s beliefs was probably right, but he was
wrongtodismissthebuyingpriceasirrelevant.Thepurchasepricedoesmatterandshould
beconsidered,evenbyEcons.Thedispositioneffectisacostlybiasbecausethequestion
of whether to sell winners or losers has a clear answer, and it is not that it makes no
difference.Ifyoucareaboutyourwealthratherthanyourimmediateemotions,youwill
selltheloserTiffanyMotorsandhangontothewinningBlueberryTiles.Atleastinthe
UnitedStates,taxesprovideastrongincentive:realizinglossesreducesyourtaxes,while
selling winners exposes you to taxes. This elementary fact of financial life is actually
knowntoallAmericaninvestors,and itdeterminesthedecisionstheymakeduringone
monthoftheyear—investorssellmorelosersinDecember,whentaxesareontheirmind.
Thetaxadvantageisavailableallyear,ofcourse,butfor11monthsoftheyearmental
accounting prevails over financial common sense. Another argument against selling
winnersisthewell-documentedmarketanomalythatstocksthatrecentlygainedinvalue
arelikelytogoongainingatleastforashortwhile.Theneteffectislarge:theexpected
after-taxextrareturnofsellingTiffanyratherthanBlueberryis3.4%overthenextyear.Cl
BTh5ingeliosingamentalaccountwithagainisapleasure,butitisapleasureyoupay
for.ThemistakeisnotonethatanEconwouldevermake,andexperiencedinvestors,who
areusingtheirSystem2,arelesssusceptibletoitthanarenovices.
A rational decision maker is interested only in the future consequences of current
investments.JustifyingearliermistakesisnotamongtheEcon’sconcerns.Thedecisionto
investadditionalresourcesinalosingaccount,whenbetterinvestmentsareavailable,is
knownasthesunk-costfallacy,acostly mistake thatisobservedindecisions large and
small.Drivingintotheblizzardbecauseonepaidforticketsisasunk-costerror.
Imagine a company that has already spent $50 million on a project. The project is
nowbehindscheduleandtheforecastsofitsultimatereturnsarelessfavorablethanatthe
initial planning stage. An additional investment of $60 million is required to give the
projectachance.Analternativeproposalistoinvestthesameamountinanewproject
that currently looks likely to bring higher returns. What will the company do? All too
often a company afflicted by sunk costs drives into the blizzard, throwing good money
afterbadratherthanacceptingthehumiliationofclosingtheaccountofacostlyfailure.
Thissituationisinthetop-rightcellofthefourfoldpattern,wherethechoiceisbetweena
surelossandanunfavorablegamble,whichisoftenunwiselypreferred.
Theescalationofcommitmenttofailingendeavorsisamistakefromtheperspective
of the firm but not necessarily from the perspective of the executive who “owns” a
flounderingproject.Cancelingtheprojectwillleaveapermanentstainontheexecutive’s
record, and his personal interests are perhaps best served by gambling further with the
organization’sresourcesinthehopeofrecoupingtheoriginalinvestment—oratleastinan
attemptto postponethe day ofreckoning. In the presence ofsunk costs, themanager’s
incentivesaremisalignedwiththeobjectivesofthefirmanditsshareholders,afamiliar
typeofwhatisknownastheagencyproblem.Boardsofdirectorsarewellawareofthese
conflictsandoftenreplaceaCEOwhoisencumberedbypriordecisionsandreluctantto
cutlosses.ThemembersoftheboarddonotnecessarilybelievethatthenewCEOismore
competent than the one she replaces. They do know that she does not carry the same
mentalaccountsandisthereforebetterabletoignorethesunkcostsofpastinvestmentsin
evaluatingcurrentopportunities.
Thesunk-costfallacykeepspeoplefortoolonginpoorjobs,unhappymarriages,and
unpromising research projects. I have often observed young scientists struggling to
salvageadoomedprojectwhentheywouldbebetteradvisedtodropitandstartanew
one. Fortunately, research suggests that at least in some contexts the fallacy can be
overcome.Thesunk-costfallacyisidentifiedandtaughtasamistakeinbotheconomics
andbusinesscourses,apparentlytogoodeffect:thereisevidencethatgraduatestudentsin
thesefieldsaremorewillingthanotherstowalkawayfromafailingproject.
Regret
Regretisanemotion,anditisalsoapunishmentthatweadministertoourselves.Thefear
ofregretisafactorinmanyofthedecisionsthatpeoplemake(“Don’tdothis,youwill
regret it” is a common warning), and the actual experience of regret is familiar. The
emotionalstatehasbeenwelldescribedbytwoDutchpsychologists,whonotedthatregret
is“accompaniedbyfeelingsthatoneshouldhaveknownbetter,byaBTh5=“4ncesinking
feeling, by thoughts about the mistake one has made and the opportunities lost, by a
tendencytokickoneselfandtocorrectone’smistake,andbywantingtoundotheevent
andtogeta second chance.”Intenseregretis what youexperiencewhenyou can most
easilyimagineyourselfdoingsomethingotherthanwhatyoudid.
Regretisoneofthecounterfactualemotionsthataretriggeredbytheavailabilityof
alternativestoreality. Aftereveryplanecrash therearespecialstories about passengers
who“shouldnot”havebeenontheplane—theygotaseatatthelastmoment,theywere
transferredfromanother airline,theywere supposedtofly aday earlier buthad had to
postpone. The common feature of these poignant stories is that they involve unusual
events—and unusual events are easier than normal events to undo in imagination.
Associative memory contains a representation of the normal world and its rules. An
abnormal event attracts attention, and it also activates the idea of the event that would
havebeennormalunderthesamecircumstances.
Toappreciatethelinkofregrettonormality,considerthefollowingscenario:
Mr.Brownalmostneverpicksuphitchhikers.Yesterdayhegavea manarideand
wasrobbed.
Mr.Smithfrequentlypicksuphitchhikers.Yesterdayhegaveamanarideandwas
robbed.
Whoofthetwowillexperiencegreaterregretovertheepisode?
Theresultsarenotsurprising:88%ofrespondentssaidMr.Brown,12%saidMr.Smith.
Regretisnotthesameasblame.Otherparticipantswereaskedthisquestionaboutthe
sameincident:
Whowillbecriticizedmostseverelybyothers?
Theresults:Mr.Brown23%,Mr.Smith77%.
Regretandblamearebothevokedbyacomparisontoanorm,buttherelevantnorms
aredifferent.TheemotionsexperiencedbyMr.BrownandMr.Smitharedominatedby
whattheyusuallydoabouthitchhikers.TakingahitchhikerisanabnormaleventforMr.
Brown, and most people therefore expect him to experience more intense regret. A
judgmental observer, however, will compare both men to conventional norms of
reasonablebehaviorandislikelytoblameMr.Smithforhabituallytakingunreasonable
risks. We are tempted to say that Mr. Smith deserved his fate and that Mr. Brown was
unlucky.ButMr.Brownistheonewhoismorelikelytobekickinghimself,becausehe
actedoutofcharacterinthisoneinstance.
Decision makers know that they are prone to regret, and the anticipation of that
painful emotion plays a part in many decisions. Intuitions about regret are remarkably
uniformandcompelling,asthenextexampleillustrates.
Paul owns shares in company A. During the past year he considered switching to
stock in company B, but he decided against it. He now learns that he would have
beenbetteroffby$1,200ifhehadswitchedtothestockofcompanyB.
GeorgeownedsharesincompanyB.DuringthepastyearheswBTh5ne
Whofeelsgreaterregret?
Theresultsareclear-cut:8%ofrespondentssayPaul,92%sayGeorge.
Thisiscurious,becausethesituationsofthetwoinvestorsareobjectivelyidentical.
TheybothnowownstockAandbothwouldhavebeenbetteroffbythesameamountif
they owned stock B. The only difference is that George got to where he is by acting,
whereasPaulgottothesameplacebyfailingtoact.Thisshortexampleillustratesabroad
story:peopleexpecttohavestrongeremotionalreactions(includingregret)toanoutcome
thatisproducedbyactionthantothesameoutcomewhenitisproducedbyinaction.This
hasbeenverifiedinthecontextofgambling:peopleexpecttobehappieriftheygamble
andwinthaniftheyrefrainfromgamblingandgetthesameamount.Theasymmetryisat
leastasstrongforlosses,anditappliestoblameaswellastoregret.Thekeyisnotthe
differencebetweencommissionandomissionbutthedistinctionbetweendefaultoptions
and actions that deviate from the default. When you deviate from the default, you can
easily imagine the norm—and if the default is associated with bad consequences, the
discrepancybetween thetwo canbe thesource of painful emotions.The defaultoption
when you own a stock is not to sell it, but the default option when you meet your
colleagueinthemorningistogreethim.Sellingastockandfailingtogreetyourcoworker
arebothdeparturesfromthedefaultoptionandnaturalcandidatesforregretorblame.
Inacompellingdemonstrationofthepowerofdefaultoptions,participantsplayeda
computersimulationofblackjack.Someplayerswereasked“Doyouwishtohit?”while
otherswere asked“Do you wishto stand?”Regardless ofthe question, sayingyes was
associatedwithmuchmoreregretthansayingnoiftheoutcomewasbad!Thequestion
evidentlysuggestsadefaultresponse,whichis,“Idon’thaveastrongwishtodoit.”Itis
thedeparturefromthedefaultthatproducesregret.Anothersituationinwhichactionis
the default is that of a coach whose team lost badly in their last game. The coach is
expectedtomakeachangeofpersonnelorstrategy,andafailuretodosowillproduce
blameandregret.
Theasymmetryintheriskofregretfavorsconventionalandrisk-aversechoices.The
biasappearsinmanycontexts.Consumerswhoareremindedthattheymayfeelregretasa
result of their choices show an increased preference for conventional options, favoring
brandnamesovergenerics.Thebehaviorofthemanagersoffinancialfundsastheyear
approachesitsendalsoshowsaneffectof anticipated evaluation: they tendtocleanup
their portfolios of unconventional and otherwise questionable stocks. Even life-or-death
decisionscanbeaffected.Imagineaphysicianwithagravelyillpatient.Onetreatmentfits
thenormalstandardofcare;anotherisunusual.Thephysicianhassomereasontobelieve
that the unconventional treatment improves the patient’s chances, but the evidence is
inconclusive.Thephysicianwhoprescribestheunusualtreatmentfacesasubstantialrisk
of regret, blame, and perhaps litigation. In hindsight, it will be easier to imagine the
normal choice; the abnormal choice will be easy to undo. True, a good outcome will
contributetothereputationofthephysicianwhodared,butthepotentialbenefitissmaller
thanthepotentialcostbecausesuccessisgenerallyamorenormaloutcomethanisfailure.
ResponsibBTh5onchepotenility
Losses are weighted about twice as much as gains in several contexts: choice between
gambles, the endowment effect, and reactions to price changes. The loss-aversion
coefficientismuchhigherinsomesituations.Inparticular,youmaybemorelossaverse
foraspectsofyourlifethataremoreimportantthanmoney,suchashealth.Furthermore,
your reluctance to “sell” important endowments increases dramatically when doing so
might make you responsible for an awful outcome. Richard Thaler’s early classic on
consumer behavior included a compelling example, slightly modified in the following
question:
Youhavebeenexposedtoadiseasewhichifcontractedleadstoaquickandpainless
deathwithinaweek.Theprobabilitythatyouhavethediseaseis1/1,000.Thereisa
vaccinethatiseffectiveonlybeforeanysymptomsappear.Whatisthemaximumyou
wouldbewillingtopayforthevaccine?
Mostpeoplearewillingtopayasignificantbutlimitedamount.Facingthepossibilityof
death is unpleasant, but the risk is small and it seems unreasonable to ruin yourself to
avoidit.Nowconsideraslightvariation:
Volunteersareneededforresearchontheabovedisease.Allthatisrequiredisthat
you expose yourself to a 1/1,000 chance of contracting the disease. What is the
minimum you would ask to be paid in order to volunteer for this program? (You
wouldnotbeallowedtopurchasethevaccine.)
As you might expect, the fee that volunteers set is far higher than the price they were
willingtopayforthevaccine.Thalerreportedinformallythatatypicalratioisabout50:1.
Theextremelyhighsellingpricereflectstwofeaturesofthisproblem.Inthefirstplace,
youarenotsupposedtosellyourhealth;thetransactionisnotconsideredlegitimateand
thereluctancetoengageinitisexpressedinahigherprice.Perhapsmostimportant,you
willberesponsiblefortheoutcomeifitisbad.Youknowthatifyouwakeuponemorning
with symptoms indicating that you will soon be dead, you will feel more regret in the
second case than in the first, because you could have rejected the idea of selling your
healthwithoutevenstoppingtoconsidertheprice.Youcouldhavestayedwiththedefault
optionanddonenothing,andnowthiscounterfactualwillhauntyoufortherestofyour
life.
The survey of parents’ reactions to a potentially hazardous insecticide mentioned
earlier also included a question about the willingness to accept increased risk. The
respondentsweretoldtoimaginethattheyusedaninsecticidewheretheriskofinhalation
andchildpoisoningwas15per10,000bottles.Alessexpensiveinsecticidewasavailable,
forwhichtheriskrosefrom15to16per10,000bottles.Theparentswereaskedforthe
discountthatwouldinducethemtoswitchtothelessexpensive(andlesssafe)product.
Morethantwo-thirdsoftheparentsinthesurveyrespondedthattheywouldnotpurchase
thenewproductatanyprice!Theywereevidentlyrevoltedbytheveryideaoftradingthe
safety of their child for money. The minority who found a discount they could accept
demandedanamountthatwassignificantlyhigherthantheamounttheywerewillingto
payforafarlargerimprovementinthesafetyoftheproduct.
Anyonecanunderstandandsympathizewiththereluctanceofparentstotradeevena
minute increase of risk to their child for money. It is worth noting, however, that this
attitudeisincoherentandpotentiallydamagingtothesafetyoftBTh5rytanceofhosewe
wishtoprotect.Eventhemostlovingparentshavefiniteresourcesoftimeandmoneyto
protecttheirchild(thekeeping-my-child-safementalaccounthasalimitedbudget),andit
seemsreasonabletodeploytheseresourcesinaway thatputsthemtobestuse.Money
thatcouldbesavedbyacceptingaminuteincreaseintheriskofharmfromapesticide
could certainly be put to better use in reducing the child’s exposure to other harms,
perhapsbypurchasingasafercarseatorcoversforelectricsockets.Thetabootradeoff
againstacceptinganyincreaseinriskisnotanefficientwaytousethesafetybudget.In
fact,theresistancemaybemotivatedbyaselfishfearofregretmorethanbyawishto
optimize the child’s safety. The what-if? thought that occurs to any parent who
deliberatelymakessuchatradeisanimageoftheregretandshameheorshewouldfeel
intheeventthepesticidecausedharm.
Theintenseaversiontotradingincreasedriskforsomeotheradvantageplaysoutona
grandscaleinthelawsandregulationsgoverningrisk.Thistrendisespeciallystrongin
Europe, where the precautionary principle, which prohibits any action that might cause
harm,isawidelyaccepteddoctrine.Intheregulatorycontext,theprecautionaryprinciple
imposestheentireburdenofprovingsafetyonanyonewhoundertakesactionsthatmight
harm people or the environment. Multiple international bodies have specified that the
absenceofscientificevidenceofpotentialdamageisnotsufficientjustificationfortaking
risks. As the jurist Cass Sunstein points out, the precautionary principle is costly, and
when interpreted strictly it can be paralyzing. He mentions an impressive list of
innovations that would not have passed the test, including “airplanes, air conditioning,
antibiotics, automobiles, chlorine, the measles vaccine, open-heart surgery, radio,
refrigeration, smallpox vaccine, and X-rays.” The strong version of the precautionary
principleisobviouslyuntenable.Butenhancedlossaversionisembeddedinastrongand
widelysharedmoralintuition;itoriginatesinSystem1.Thedilemmabetweenintensely
loss-averse moral attitudes and efficient risk management does not have a simple and
compellingsolution.
Wespendmuchofourdayanticipating,andtryingtoavoid,theemotionalpainsweinflict
on ourselves. How seriously should we take these intangible outcomes, the self-
administered punishments (and occasional rewards) that we experience as we scoreour
lives?Econsarenotsupposedtohavethem,andtheyarecostlytoHumans.Theyleadto
actionsthataredetrimentaltothewealthofindividuals,tothesoundnessofpolicy,andto
thewelfareofsociety.Buttheemotionsofregretandmoralresponsibilityarereal,andthe
factthatEconsdonothavethemmaynotberelevant.
Isitreasonable,inparticular,toletyourchoicesbeinfluencedbytheanticipationof
regret? Susceptibility to regret, like susceptibility to fainting spells, is a fact of life to
whichonemustadjust.Ifyouareaninvestor,sufficientlyrichandcautiousatheart,you
maybeableto affordtheluxuryof a portfoliothatminimizestheexpectationof regret
evenifitdoesnotmaximizetheaccrualofwealth.
Youcanalsotakeprecautionsthatwillinoculateyouagainstregret.Perhapsthemost
usefulistobeexplicitabouttheanticipationofregret.Ifyoucanrememberwhenthings
gobadlythatyouconsideredthepossibilityofregretcarefullybeforedeciding,youare
likelytoexperiencelessof it.Youshouldalsoknowthat regretandhindsightbiaswill
cometogether,soanythingyoucandotoprecludehindsightislikelytobehelpful.My
personal hindsight-avoiding B Th5he ything policy is to be either very thorough or
completely casual when making a decision with long-term consequences. Hindsight is
worsewhenyouthinkalittle,justenoughtotellyourselflater,“Ialmostmadeabetter
choice.”
DanielGilbertandhiscolleaguesprovocativelyclaimthatpeoplegenerallyanticipate
moreregretthantheywillactuallyexperience,becausetheyunderestimatetheefficacyof
thepsychologicaldefensestheywilldeploy—whichtheylabelthe“psychologicalimmune
system.” Their recommendation is that you should not put too much weight on regret;
evenifyouhavesome,itwillhurtlessthanyounowthink.
SpeakingofKeepingScore
“Hehasseparatementalaccountsforcashandcreditpurchases.Iconstantlyremind
himthatmoneyismoney.”
“Wearehangingontothatstockjusttoavoidclosingourmentalaccountataloss.
It’sthedispositioneffect.”
“Wediscoveredanexcellentdishatthatrestaurantandwenevertryanythingelse,to
avoidregret.”
“Thesalespersonshowedmethemostexpensivecarseatandsaiditwasthesafest,
andIcouldnotbringmyselftobuythecheapermodel.Itfeltlikeatabootradeoff.”
P
Reversals
Youhavethetaskofsettingcompensationforvictimsofviolentcrimes.Youconsider
thecaseofamanwholosttheuseofhisrightarmasaresultofagunshotwound.He
was shot when he walked in on a robbery occurring in a convenience store in his
neighborhood.
Twostoreswere located near thevictim’shome,oneof which hefrequentedmore
regularlythantheother.Considertwoscenarios:
(i)Theburglaryhappenedintheman’sregularstore.
(ii) The man’s regular store was closed for a funeral, so he did his shopping in the
otherstore,wherehewasshot.
Shouldthestoreinwhichthemanwasshotmakeadifferencetohiscompensation?
You made your judgment in joint evaluation, where you consider two scenarios at the
same time and make a comparison. You can apply a rule. If you think that the second
scenariodeserveshighercompensation,youshouldassignitahigherdollarvalue.
Thereisalmostuniversalagreementontheanswer:compensationshouldbethesame
inbothsituations.Thecompensationisforthecripplinginjury,sowhyshouldthelocation
inwhichitoccurredmakeanydiffCmakerence?Thejointevaluationofthetwoscenarios
gaveyouachancetoexamineyourmoralprinciplesaboutthefactorsthatarerelevantto
victimcompensation. Formost people, location is notone of these factors. Asin other
situations that require an explicit comparison, thinking was slow and System 2 was
involved.
ThepsychologistsDaleMillerandCathyMcFarland,whooriginallydesignedthetwo
scenarios, presented them to different people for single evaluation. In their between-
subjectsexperiment,eachparticipantsawonlyonescenarioandassignedadollarvalueto
it.Theyfound,asyousurelyguessed,thatthevictimwasawardedamuchlargersumifhe
wasshotinastoreherarelyvisitedthanifhewasshotinhisregularstore.Poignancy(a
closecousinofregret)isacounterfactualfeeling,whichisevokedbecausethethought“if
onlyhehadshoppedathisregularstore…”comesreadilytomind.ThefamiliarSystem1
mechanismsofsubstitutionandintensitymatchingtranslatethestrengthoftheemotional
reactiontothestoryontoamonetaryscale,creatingalargedifferenceindollarawards.
The comparison of the two experiments reveals a sharp contrast. Almost everyone
whoseesbothscenariostogether(within-subject)endorsestheprinciplethatpoignancyis
nota legitimateconsideration.Unfortunately,the principle becomesrelevant onlywhen
thetwoscenariosareseentogether,andthisisnothowlifeusuallyworks.Wenormally
experiencelifeinthebetween-subjectsmode,inwhichcontrastingalternativesthatmight
changeyourmindareabsent,andofcourseWYSIATI.Asaconsequence,thebeliefsthat
you endorse when you reflect about morality do not necessarily govern youremotional
reactions,andthemoralintuitionsthatcometoyourmindindifferentsituationsarenot
internallyconsistent.
Thediscrepancybetweensingleandjointevaluationoftheburglaryscenariobelongs
toabroadfamilyofreversalsofjudgmentandchoice.Thefirstpreferencereversalswere
discoveredintheearly1970s,andmanyreversalsofotherkindswerereportedoverthe
years.
ChallengingEconomics
Preferencereversalshaveanimportantplaceinthehistoryoftheconversationbetween
psychologists and economists. The reversals that attracted attention were reported by
SarahLichtensteinandPaulSlovic,twopsychologistswhohaddonetheirgraduatework
attheUniversityofMichiganatthesametimeasAmos.Theyconductedanexperiment
onpreferencesbetweenbets,whichIshowinaslightlysimplifiedversion.
You are offered a choice between two bets, which are to be played on a roulette
wheelwith36sectors.
BetA:11/36towin$160,25/36tolose$15
BetB:35/36towin$40,1/36tolose$10
Youareaskedtochoosebetweenasafebetandariskierone:analmostcertainwinofa
modest amount, or a small chance to win a substantially larger amount and a high
probabilityoflosing.Safetyprevails,andBisclearlythemorepopularchoice.
Nowconsidereachbetseparately:Ifyouownedthatbet,whatisthelowestpriceat
whichyouwouldsellit?Rememberthatyouarenotnegotiatingwithanyone—yourtask
istodeterminethelowestpriceatwhichyouwouldtrulybewillingtogiveupthebet.Try
it.YoumayfindthattheprizethatcanbewonisBmaktwearenotsalientinthistask,and
thatyourevaluationofwhatthebetisworthisanchoredonthatvalue.Theresultssupport
thisconjecture,andthesellingpriceishigherforbetAthanforbetB.Thisisapreference
reversal:peoplechooseBoverA,butiftheyimagineowningonlyoneofthem,theyseta
highervalueonAthanonB.Asintheburglaryscenarios,thepreferencereversaloccurs
becausejointevaluationfocusesattentiononanaspectofthesituation—thefactthatbetA
ismuchlesssafethanbetB—whichwaslesssalientinsingleevaluation.Thefeaturesthat
caused the difference between the judgments of the options in single evaluation—the
poignancyofthevictimbeinginthewronggrocerystoreandtheanchoringontheprize—
are suppressed or irrelevant when the options are evaluated jointly. The emotional
reactions of System 1 are much more likely to determine single evaluation; the
comparison that occurs in joint evaluation always involves a more careful and effortful
assessment,whichcallsforSystem2.
The preference reversal can be confirmed in a within-subject experiment, in which
subjects set prices on both sets as part of a long list, and also choose between them.
Participantsareunawareoftheinconsistency,andtheirreactionswhenconfrontedwithit
can be entertaining. A 1968 interview of a participant in the experiment, conducted by
Sarah Lichtenstein, is an enduring classic of the field. The experimenter talks at length
withabewilderedparticipant,whochoosesonebetoveranotherbutisthenwillingtopay
moneytoexchangetheitemhejustchosefortheonehejustrejected,andgoesthrough
thecyclerepeatedly.
Rational Econs would surely not be susceptible to preference reversals, and the
phenomenonwasthereforeachallengetotherational-agentmodelandtotheeconomic
theorythatisbuiltonthismodel.Thechallengecouldhavebeenignored,butitwasnot.A
fewyearsafterthepreferencereversalswerereported,tworespectedeconomists,David
Grether and Charles Plott, published an article in the prestigious American Economic
Review,inwhichtheyreportedtheirownstudiesofthephenomenonthatLichtensteinand
Slovichaddescribed.Thiswasprobablythefirstfindingbyexperimentalpsychologists
thateverattractedtheattentionofeconomists.TheintroductoryparagraphofGretherand
Plott’sarticlewasunusuallydramaticforascholarlypaper,andtheirintentwasclear:“A
body of data and theory has been developing within psychology which should be of
interesttoeconomists.Takenatfacevaluethedataaresimplyinconsistentwithpreference
theory and have broad implications about research priorities within economics…. This
paperreportstheresultsofaseriesofexperimentsdesignedtodiscreditthepsychologists’
worksasappliedtoeconomics.”
GretherandPlottlistedthirteentheoriesthatcouldexplaintheoriginalfindingsand
reported carefully designed experiments that tested these theories. One of their
hypotheses,which—needlesstosay—psychologistsfoundpatronizing,wasthattheresults
were due to the experiment being carried out by psychologists! Eventually, only one
hypothesis was left standing: the psychologists were right. Grether and Plott
acknowledged that this hypothesis is the least satisfactory from the point of view of
standardpreferencetheory,because“itallowsindividualchoicetodependonthecontext
inwhichthechoicesaremade”—aclearviolationofthecoherencedoctrine.
You might think that this surprising outcome would cause much anguished soul-
searchingamongeconomists,asabasicassumptionoftheirtheoryhadbeensuccessfully
challenged.Butthisisnotthewaythingsworkinsocialscience,includingbothpsychol
Bmak/p>ished soogy and economics. Theoretical beliefs are robust, and it takes much
morethanoneembarrassingfindingforestablishedtheoriestobeseriouslyquestioned.In
fact, Grether and Plott’s admirably forthright report had little direct effect on the
convictionsofeconomists,probablyincludingGretherandPlott.Itcontributed,however,
toa greaterwillingnessof thecommunity ofeconomiststo takepsychological research
seriously and thereby greatly advanced the conversation across the boundaries of the
disciplines.
Categories
“HowtallisJohn?”IfJohnis5’tall,youranswerwilldependonhisage;heisverytallif
heis6yearsold,veryshortifheis16.YourSystem1automaticallyretrievestherelevant
norm,andthemeaningofthescaleoftallnessisadjustedautomatically.Youarealsoable
to match intensities across categories and answer the question, “How expensive is a
restaurantmealthat matchesJohn’sheight?” Your answerwilldepend onJohn’s age: a
muchlessexpensivemealifheis16thanifheis6.
Butnowlookatthis:
Johnis6.Heis5’tall.
Jimis16.Heis5′1″tall.
Insingleevaluations,everyonewill agree thatJohnisvery tall andJimisnot, because
theyarecomparedtodifferentnorms.Ifyouareaskedadirectlycomparativequestion,“Is
JohnastallasJim?”youwillanswerthatheisnot.Thereisnosurprisehereandlittle
ambiguity.Inothersituations,however,theprocessbywhichobjectsandeventsrecruit
theirowncontextofcomparisoncanleadtoincoherentchoicesonseriousmatters.
You should not form the impression that single and joint evaluations are always
inconsistent,orthatjudgmentsarecompletelychaotic.Ourworldisbrokenintocategories
forwhichwehavenorms,suchassix-year-oldboysortables.Judgmentsandpreferences
are coherent within categories but potentially incoherent when the objects that are
evaluated belong to different categories. For an example, answer the following three
questions:
Whichdoyoulikemore,applesorpeaches?
Whichdoyoulikemore,steakorstew?
Whichdoyoulikemore,applesorsteak?
Thefirstandthesecondquestionsrefertoitemsthatbelongtothesamecategory,andyou
knowimmediatelywhichyoulikemore.Furthermore,youwouldhaverecoveredthesame
rankingfromsingleevaluation(“Howmuchdoyoulikeapples?”and“Howmuchdoyou
likepeaches?”)becauseapplesandpeachesbothevokefruit.Therewillbenopreference
reversalbecausedifferentfruitsarecomparedtothesamenormandimplicitlycompared
to each other in single as well as in joint evaluation. In contrastto the within-category
questions,thereisnostableanswerforthecomparisonofapplesandsteak.Unlikeapples
and peaches, apples and steak are not natural substitutes and they do not fill the same
need.Yousometimeswantsteakandsometimesanapple,butyourarelysaythateither
onewilldojustaswellastheother.
Imaginereceivingane-mailfromanorganizationthatyougenerallytrust,requesting
aBmak
Dolphinsinmanybreedinglocationsarethreatenedbypollution,whichisexpected
toresultinadeclineofthedolphinpopulation.Aspecialfundsupportedbyprivate
contributions has been set up to provide pollution-free breeding locations for
dolphins.
Whatassociationsdidthisquestionevoke?Whetherornotyouwerefullyawareofthem,
ideasand memoriesof relatedcauses cameto yourmind. Projectsintended topreserve
endangeredspecieswereespeciallylikelytoberecalled.EvaluationontheGOOD–BAD
dimensionisanautomaticoperationofSystem1,andyouformedacrudeimpressionof
the ranking of the dolphin among the species that came to mind. The dolphin is much
morecharmingthan,say,ferrets,snails,orcarp—ithasahighlyfavorablerankintheset
ofspeciestowhichitisspontaneouslycompared.
Thequestionyoumustanswerisnotwhetheryoulikedolphinsmorethancarp;you
have been asked to come up with a dollar value. Of course, you may know from the
experienceofprevioussolicitationsthatyouneverrespondtorequestsofthiskind.Fora
fewminutes,imagineyourselfassomeonewhodoescontributetosuchappeals.
Likemanyotherdifficultquestions,theassessmentofdollarvaluecanbesolvedby
substitutionandintensitymatching.Thedollarquestionisdifficult,butaneasierquestion
isreadilyavailable.Becauseyoulikedolphins,youwillprobablyfeelthatsavingthemis
a good cause. The next step, which is also automatic, generates a dollar number by
translatingtheintensityofyourlikingofdolphinsontoascaleofcontributions.Youhave
asenseofyourscaleofpreviouscontributionstoenvironmentalcauses,whichmaydiffer
fromthescaleofyourcontributionstopoliticsortothefootballteamofyouralmamater.
Youknowwhatamountwouldbea“verylarge”contributionforyouandwhatamounts
are“large,”“modest,”and“small.”Youalsohavescalesforyourattitudetospecies(from
“likeverymuch”to“notatall”).Youarethereforeabletotranslateyourattitudeontothe
dollarscale,movingautomaticallyfrom“likealot”to“fairlylargecontribution”andfrom
theretoanumberofdollars.
Onanotheroccasion,youareapproachedwithadifferentappeal:
Farmworkers,whoareexposedtothesunformanyhours,haveahigherrateofskin
cancerthanthegeneralpopulation.Frequentmedicalcheck-upscanreducetherisk.
Afundwillbesetuptosupportmedicalcheck-upsforthreatenedgroups.
Is this an urgent problem? Which category did it evoke as a norm when you assessed
urgency? If you automatically categorized the problem as a public-health issue, you
probably found that the threat of skin cancer in farmworkers does not rank very high
amongtheseissues—almostcertainlylowerthantherankofdolphinsamongendangered
species.Asyoutranslatedyourimpressionoftherelativeimportanceoftheskincancer
issueintoadollaramount,youmightwellhavecomeupwithasmallercontributionthan
you offered to protect an endearing animal. In experiments, the dolphins attracted
somewhatlargercontributionsinsingleevaluationthandidthefarmworkers.
Next, consider the two causes in joint evaluation. Which of the two, dolphins or
farmworkers, deserves a larger dollar contribution? Joint evaluation highlights a feature
that was not noticeable in si Bmakecksider the ngle evaluation but is recognized as
decisivewhendetected:farmersarehuman,dolphinsarenot.Youknewthat,ofcourse,
butitwasnotrelevanttothejudgmentthatyoumadeinsingleevaluation.Thefactthat
dolphins are not human did not arise because all the issues that were activated in your
memorysharedthatfeature.Thefactthatfarmworkersarehumandidnotcometomind
becauseallpublic-healthissuesinvolvehumans.Thenarrowframingofsingleevaluation
alloweddolphinstohaveahigherintensityscore,leadingtoahighrateofcontributions
by intensity matching. Joint evaluation changes the representation of the issues: the
“humanvs.animal”featurebecomessalientonlywhenthetwoareseentogether.Injoint
evaluation people show a solid preference for the farmworkers and a willingness to
contributesubstantiallymoretotheirwelfarethantotheprotectionofalikablenon-human
species.Hereagain,asinthecasesofthebetsandtheburglaryshooting,thejudgments
madeinsingleandinjointevaluationwillnotbeconsistent.
Christopher Hsee, of the University of Chicago, has contributed the following
exampleofpreferencereversal,amongmanyothersofthesametype.Theobjectstobe
evaluatedaresecondhandmusicdictionaries.
DictionaryA DictionaryB
Yearofpublication 1993 1993
Numberofentries 10,000 20,000
Condition Likenew Covertorn,otherwiselikenew
When the dictionaries are presented in single evaluation, dictionary A is valued more
highly, but of course the preference changes in joint evaluation. The result illustrates
Hsee’s evaluability hypothesis: The number of entries is given no weight in single
evaluation,becausethenumbersarenot“evaluable”ontheirown.Injointevaluation,in
contrast,itisimmediatelyobviousthatdictionaryBissuperioronthisattribute,anditis
alsoapparentthatthenumberofentriesisfarmoreimportantthantheconditionofthe
cover.
UnjustReversals
Thereisgoodreasontobelievethattheadministrationofjusticeisinfectedbypredictable
incoherence in several domains. The evidence is drawn in part from experiments,
includingstudiesofmockjuries,andinpartfromobservationofpatternsinlegislation,
regulation,andlitigation.
Inoneexperiment,mockjurorsrecruitedfromjuryrollsinTexaswereaskedtoassess
punitivedamagesinseveralcivilcases.Thecasescameinpairs,eachconsistingofone
claimforphysicalinjuryandoneforfinancialloss.Themockjurorsfirstassessedoneof
thescenariosandthentheywereshownthecasewithwhichitwasBmakin,eacpaired
andwereaskedtocomparethetwo.Thefollowingaresummariesofonepairofcases:
Case 1: A child suffered moderate burns when his pajamas caught fire as he was
playing with matches. The firm that produced the pajamas had not made them
adequatelyfireresistant.
Case 2: The unscrupulous dealings of a bank caused another bank a loss of $10
million.
Halfoftheparticipantsjudgedcase1first(insingleevaluation)beforecomparingthetwo
casesinjointevaluation.Thesequencewasreversedfortheotherparticipants.Insingle
evaluation,thejurorsawardedhigherpunitivedamagestothedefraudedbankthantothe
burnedchild,presumablybecausethesizeofthefinanciallossprovidedahighanchor.
When the cases were considered together, however, sympathy for the individual
victimprevailedovertheanchoringeffectandthejurorsincreasedtheawardtothechild
tosurpasstheawardtothebank.Averagingoverseveralsuchpairsofcases,awardsto
victimsofpersonalinjuryweremorethantwiceaslargeinjointthaninsingleevaluation.
Thejurorswhosawthecaseoftheburnedchildonitsownmadeanofferthatmatchedthe
intensity of their feelings. They could not anticipate that the award to the child would
appear inadequate in the context of a large award to a financial institution. In joint
evaluation,thepunitiveawardtothebankremainedanchoredonthelossithadsustained,
buttheawardtotheburnedchildincreased,reflectingtheoutrageevokedbynegligence
thatcausesinjurytoachild.
Aswehaveseen,rationalityisgenerallyservedbybroaderandmorecomprehensive
frames,andjointevaluationisobviouslybroaderthansingleevaluation.Ofcourse,you
shouldbewaryofjointevaluationwhensomeonewhocontrolswhatyouseehasavested
interestinwhatyouchoose.Salespeoplequicklylearnthatmanipulationofthecontextin
whichcustomersseeagoodcanprofoundlyinfluencepreferences.Exceptforsuchcases
ofdeliberatemanipulation,thereisapresumptionthatthecomparativejudgment,which
necessarilyinvolvesSystem2,ismorelikelytobestablethansingleevaluations,which
oftenreflecttheintensityofemotionalresponsesofSystem1.Wewouldexpectthatany
institutionthatwishestoelicitthoughtfuljudgmentswouldseektoprovidethejudgeswith
abroadcontextfortheassessmentsofindividualcases.IwassurprisedtolearnfromCass
Sunstein that jurors who are to assess punitive damages are explicitly prohibited from
considering other cases. The legal system, contrary to psychological common sense,
favorssingleevaluation.
In another study of incoherence in the legal system, Sunstein compared the
administrative punishments that can be imposed by different U.S. government agencies
including the Occupational Safety and Health Administration and the Environmental
Protection Agency. He concluded that “within categories, penalties seem extremely
sensible,atleastinthesensethatthemoreseriousharmsarepunishedmoreseverely.For
occupationalsafetyandhealthviolations,thelargestpenaltiesareforrepeatedviolations,
thenextlargestforviolationsthatarebothwillfulandserious,andtheleastseriousfor
failurestoengageintherequisiterecord-keeping.”Itshouldnotsurpriseyou,however,
thatthesizeofpenaltiesvariedgreatlyacrossagencies,inamannerthatreflectedpolitics
andhistorymorethananyglobalconcernforfairness.Thefinefora“seriousviolation”of
theregulationsconcerningworkersafetyiscappedat$7,000,whileaviBmaknseflected
polationoftheWildBirdConservationActcanresultinafineofupto$25,000.Thefines
aresensibleinthecontextofotherpenaltiessetbyeachagency,buttheyappearoddwhen
comparedtoeachother.Asintheotherexamplesinthischapter,youcanseetheabsurdity
only when the two cases are viewed together in a broad frame. The system of
administrativepenaltiesiscoherentwithinagenciesbutincoherentglobally.
SpeakingofReversals
“TheBTUunits meantnothingto meuntilI sawhowmuch air-conditioning units
vary.Jointevaluationwasessential.”
“You say this was an outstanding speech because you compared it to her other
speeches.Comparedtoothers,shewasstillinferior.”
“It is often the case that when you broaden the frame, you reach more reasonable
decisions.”
“When you see cases in isolation, you are likely to be guided by an emotional
reactionofSystem1.”
P
FramesandReality
ItalyandFrancecompetedin the2006 finaloftheWorldCup. The nexttwosentences
bothdescribetheoutcome:“Italywon.”“Francelost.”Dothosestatementshavethesame
meaning?Theanswerdependsentirelyonwhatyoumeanbymeaning.
Forthepurposeoflogicalreasoning,thetwodescriptionsoftheoutcomeofthematch
areinterchangeablebecausetheydesignatethesamestateoftheworld.Asphilosophers
say,theirtruthconditionsareidentical:ifoneofthesesentencesistrue,thentheotheris
true as well. This is how Econs understand things. Their beliefs and preferences are
reality-bound.Inparticular,theobjectsoftheirchoicesarestatesoftheworld,whichare
notaffectedbythewordschosentodescribethem.
There is another sense of meaning, in which “Italy won” and “France lost” do not
havethesamemeaningatall.Inthissense,themeaningofasentenceiswhathappensin
yourassociativemachinerywhileyouunderstandit.Thetwosentencesevokemarkedly
differentassociations.“Italywon”evokesthoughtsoftheItalianteamandwhatitdidto
win.“Francelost”evokesthoughtsoftheFrenchteamandwhatitdidthatcauseditto
lose,includingthememorableheadbuttofanItalianplayerbytheFrenchstarZidane.In
terms of the associations they bring to mind—how System 1 reacts to them—the two
sentences really “mean” different things. The fact that logically equivalent statements
evoke different reactions makes it impossible for Humans to be as reliably rational as
Econs.
EmotionalFraming
AmosandIappliedthelabelofframingeffectstotheunjustifiedinfluencesofformulation
onbeliefsanCondpreferences.Thisisoneoftheexamplesweused:
Wouldyouacceptagamblethatoffersa10%chancetowin$95anda90%chanceto
lose$5?
Wouldyoupay$5toparticipateinalotterythatoffersa10%chancetowin$100and
a90%chancetowinnothing?
First,takeamomenttoconvinceyourselfthatthetwoproblemsareidentical.Inbothof
themyoumustdecidewhethertoacceptanuncertainprospectthatwillleaveyoueither
richerby$95orpoorerby$5.Someonewhosepreferencesarereality-boundwouldgive
the same answer to both questions, but such individuals are rare. In fact, one version
attractsmanymorepositiveanswers:thesecond.Abadoutcomeismuchmoreacceptable
ifitisframedasthecostofalotteryticketthatdidnotwinthanifitissimplydescribedas
losingagamble.Weshouldnotbesurprised:lossesevokesstrongernegativefeelingsthan
costs.Choicesarenotreality-boundbecauseSystem1isnotreality-bound.
Theproblem we constructedwas influencedby whatwe had learnedfrom Richard
Thaler,whotoldusthatwhenhewasagraduatestudenthehadpinnedonhisboardacard
thatsaidcostsarenotlosses.Inhisearlyessayonconsumerbehavior,Thalerdescribed
the debate about whether gas stations would be allowed to charge different prices for
purchases paid with cash or on credit. The credit-card lobby pushed hard to make
differentialpricingillegal,butithadafallbackposition:thedifference,ifallowed,would
belabeleda cashdiscount,not acreditsurcharge.Theirpsychology was sound:people
willmorereadilyforgoadiscountthanpaya surcharge.Thetwomaybeeconomically
equivalent,buttheyarenotemotionallyequivalent.
In an elegant experiment, a team of neuroscientists at University College London
combinedastudyofframingeffectswithrecordingsofactivityindifferentareasofthe
brain.Inordertoprovidereliablemeasuresofthebrainresponse,theexperimentconsisted
ofmanytrials.Figure14illustratesthetwostagesofoneofthesetrials.
First,thesubjectisaskedtoimaginethatshereceivedanamountofmoney,inthis
example£50.
Thesubjectisthenaskedtochoosebetweenasureoutcomeandagambleonawheel
ofchance.Ifthewheelstopsonwhiteshe“receives”theentireamount;ifitstopsonblack
shegets nothing. Thesure outcome issimply the expectedvalue of thegamble, in this
caseagainof£20.
Figure14
Asshown,thesamesureoutcomecanbeframedintwodifferentways:asKEEP£20
orasLOSE£30.Theobjectiveoutcomesarepreciselyidenticalinthetwoframes,anda
reality-bound Econ would respond to both in the same way—selecting either the sure
thingorthegambleregardlessoftheframe—butwealreadyknowthattheHumanmindis
notboundtoreality.Tendenciestoapproachoravoidareevokedbythewords,andwe
expectSystem1tobebiasedinfavorofthesureoptionwhenitisdesignatedasKEEP
andagainstthatsameoptionwhenitisdesignatedasLOSE.
Theexperimentconsistedofmanytrials,andeachparticipantencountereBonp>
Theactivityofthebrainwasrecordedasthesubjectsmadeeachdecision.Later,the
trialswereseparatedintotwocategories:
1Trialsonwhichthesubject’schoiceconformedtotheframe
preferredthesurethingintheKEEPversion
preferredthegambleintheLOSSversion
2Trialsinwhichthechoicedidnotconformtotheframe.
Theremarkableresultsillustratethepotentialofthenewdisciplineofneuroeconomics—
thestudyofwhata person’s brain does while he makesdecisions.Neuroscientistshave
runthousandsofsuchexperiments,andtheyhavelearnedtoexpectparticularregionsof
thebrainto“lightup”—indicatingincreasedflowofoxygen,whichsuggestsheightened
neuralactivity—dependingonthenatureofthetask.Differentregionsareactivewhenthe
individualattendstoavisualobject,imagineskickingaball,recognizesaface,orthinks
of a house. Other regions light up when the individual is emotionally aroused, is in
conflict,or concentrates onsolving aproblem. Althoughneuroscientistscarefully avoid
thelanguageof“thispartofthebraindoessuchandsuch…,”theyhavelearnedagreat
dealaboutthe“personalities”ofdifferentbrainregions,andthecontributionofanalyses
ofbrainactivitytopsychologicalinterpretationhasgreatlyimproved.Theframingstudy
yieldedthreemainfindings:
A region that is commonly associated with emotional arousal (the amygdala) was
mostlikelytobeactivewhensubjects’choicesconformedtotheframe.Thisisjustas
we would expect if the emotionally loaded words KEEP and LOSE produce an
immediatetendencytoapproachthesurething(whenitisframedasagain)oravoid
it(whenitisframedasaloss).Theamygdalaisaccessedveryrapidlybyemotional
stimuli—anditisalikelysuspectforinvolvementinSystem1.
A brain region known to be associated with conflict and self-control (the anterior
cingulate) was more active when subjects did not do what comes naturally—when
theychosethesurethinginspiteofitsbeinglabeledLOSE.Resistingtheinclination
ofSystem1apparentlyinvolvesconflict.
Themost“rational”subjects—thosewhoweretheleastsusceptibletoframingeffects
—showed enhanced activity in a frontal area of the brain that is implicated in
combining emotion and reasoning to guide decisions. Remarkably, the “rational”
individualswerenotthosewhoshowedthestrongestneuralevidenceofconflict.It
appearsthattheseeliteparticipantswere(often,notalways)reality-boundwithlittle
conflict.
Byjoiningobservationsofactualchoiceswithamappingofneuralactivity,thisstudy
providesagoodillustrationofhowtheemotionevokedbyawordcan“leak”intothefinal
choice.
AnexperimentthatAmoscarriedoutwithcolleaguesatHarvardMedicalSchoolis
the classic example of emotional framing. Physician participants were given statistics
abouttheoutcomesoftwotreatmentsforlungcancer:surgeryandradiation.Thefive-year
survivalratesclearlyfavorsurgery,butintheshorttermsurgeryisriskierthanradiation.
Half the participants read statistics about survival rates, the others received the same
informationintermsofmortalityrates.Thetwodescriptionsoftheshort-termoutcomes
ofsurgerywere:
Theone-monthsurvivalrateis90%.
Thereis10%mortalityinthefirstmonth.
Youalreadyknowtheresults:surgerywasmuchmorepopularintheformerframe(84%
of physicians chose it) than in the latter (where 50% favored radiation). The logical
equivalence of the two descriptions is transparent, and a reality-bound decision maker
wouldmakethesamechoiceregardlessofwhichversionshesaw.ButSystem1,aswe
havegottentoknowit,israrelyindifferenttoemotionalwords:mortalityisbad,survival
isgood,and90%survivalsoundsencouragingwhereas10%mortalityisfrightening.An
importantfindingofthestudyisthatphysicianswerejustassusceptibletotheframing
effect as medically unsophisticated people (hospital patients and graduate students in a
businessschool).Medicaltrainingis,evidently,nodefenseagainstthepowerofframing.
The KEEP–LOSE study and the survival–mortality experiment differed in one
importantrespect.Theparticipantsin thebrain-imagingstudyhad many trialsinwhich
they encountered the different frames. They had an opportunity to recognize the
distractingeffectsoftheframesandtosimplifytheirtaskbyadoptingacommonframe,
perhaps by translating the LOSE amount into its KEEP equivalent. It would take an
intelligentperson(andanalertSystem2)tolearntodothis,andthefewparticipantswho
managed the feat were probably among the “rational” agents that the experimenters
identified.Incontrast,thephysicianswhoreadthestatisticsaboutthetwotherapiesinthe
survivalframehadnoreasontosuspectthattheywouldhavemadeadifferentchoiceif
theyhadheardthesamestatisticsframedintermsofmortality.Reframingiseffortfuland
System2isnormallylazy.Unlessthereisanobviousreasontodootherwise,mostofus
passively accept decision problems as they are framed and therefore rarely have an
opportunitytodiscovertheextenttowhichourpreferencesareframe-boundratherthan
reality-bound.
EmptyIntuitions
AmosandIintroducedourdiscussionofframingbyanexamplethathasbecomeknown
asthe“Asiandiseaseproblem”:
Imagine that the United States is preparing for the outbreak of an unusual Asian
disease,whichisexpectedtokill600people.Twoalternativeprogramstocombatthe
disease have been proposed. Assume that the exact scientific estimates of the
consequencesoftheprogramsareasfollows:
IfprogramAisadopted,200peoplewillbesaved.
IfprogramBisadopted,thereisaone-thirdprobabilitythat600peoplewillbe
savedandatwo-thirdsprobabilitythatnopeoplewillbesaved.
A substantial majority of respondents choose program A: they prefer the certain option
overthegamble.
Theoutcomesoftheprogramsareframeddifferentlyinasecondversion:
IfprogramA’isadopted,400peoplewilldie.
IfprogramB’isadopted,thereisaone-thirdprobabilitythatnobodywilldieanda
two-thirdsprobabilitythat600peoplewilldie.
Lookcloselyandcomparethetwoversions:theconsequencesofprogramsAandA’are
identical;soaretheconsequencesofprogramsBandB’.Inthesecondframe,however,a
largemajorityofpeoplechoosethegamble.
Thedifferentchoicesinthetwoframesfitprospecttheory,inwhichchoicesbetween
gamblesandsurethingsareresolveddifferently,dependingonwhethertheoutcomesare
goodorbad.Decisionmakerstendtopreferthesurethingoverthegamble(theyarerisk
averse) when the outcomes are good. They tend to reject the sure thing and accept the
gamble(theyareriskseeking)whenbothoutcomesarenegative.Theseconclusionswere
wellestablishedforchoicesaboutgamblesandsurethingsinthedomainofmoney.The
diseaseproblemshowsthatthesameruleapplieswhentheoutcomesaremeasuredinlives
savedorlost.Inthiscontext,aswell,theframingexperimentrevealsthatrisk-averseand
risk-seeking preferences are not reality-bound. Preferences between the same objective
outcomesreversewithdifferentformulations.
AnexperiencethatAmossharedwithmeaddsa grimnotetothestory.Amoswas
invitedtogiveaspeechtoagroupofpublic-healthprofessionals—thepeoplewhomake
decisions about vaccines and other programs. He took the opportunity to present them
withtheAsiandiseaseproblem:halfsawthe“lives-saved”version,theothersanswered
the “lives-lost” question. Like other people, these professionals were susceptible to the
framingeffects.Itissomewhatworryingthattheofficialswhomakedecisionsthataffect
everyone’s health can be swayed by such a superficial manipulation—but we must get
usedtotheideathatevenimportantdecisionsareinfluenced,ifnotgoverned,bySystem
1.
Even more troubling is what happens when people are confronted with their
inconsistency:“Youchosetosave200livesforsureinoneformulationandyouchoseto
gambleratherthanaccept400deathsintheother.Nowthatyouknowthesechoiceswere
inconsistent, how do you decide?” The answer is usually embarrassed silence. The
intuitionsthatdeterminedtheoriginalchoicecamefromSystem1andhadnomoremoral
basisthandidthepreferenceforkeeping£20ortheaversiontolosing£30.Savinglives
withcertaintyisgood,deathsarebad.MostpeoplefindthattheirSystem2hasnomoral
intuitionsofitsowntoanswerthequestion.
IamgratefultothegreateconomistThomasSchellingformyfavoriteexampleofa
framingeffect,whichhedescribedinhisbookChoiceandConsequence.Schelling’sbook
was written before our work on framing was published, and framing was not his main
concern.HereportedonhisexperienceteachingaclassattheKennedySchoolatHarvard,
in which Bon he linthe topic was child exemptions in the tax code. Schelling told his
studentsthatastandardexemptionisallowedforeachchild,andthattheamountofthe
exemption is independent of the taxpayer’s income. He asked their opinion of the
followingproposition:
Shouldthechildexemptionbelargerfortherichthanforthepoor?
YourownintuitionsareverylikelythesameasthoseofSchelling’sstudents:theyfound
theideaoffavoringtherichbyalargerexemptioncompletelyunacceptable.
Schellingthenpointedoutthatthetaxlawisarbitrary.Itassumesachildlessfamilyas
thedefaultcaseandreducesthetaxbytheamountoftheexemptionforeachchild.The
taxlawcouldofcourseberewrittenwithanotherdefaultcase:afamilywithtwochildren.
Inthisformulation,familieswithfewerthanthedefaultnumberofchildrenwouldpaya
surcharge.Schellingnowaskedhisstudentstoreporttheirviewofanotherproposition:
Shouldthechildlesspoorpayaslargeasurchargeasthechildlessrich?
Hereagainyouprobablyagreewiththestudents’reactiontothisidea,whichtheyrejected
withasmuchvehemenceasthefirst.ButSchellingshowedhisclassthattheycouldnot
logicallyrejectbothproposals.Setthetwoformulationsnexttoeachother.Thedifference
betweenthetaxduebyachildlessfamilyandbyafamilywithtwochildrenisdescribed
asa reductionoftaxinthefirstversionandasanincreaseinthesecond. Ifinthefirst
versionyouwantthepoortoreceivethesame(orgreater)benefitastherichforhaving
children,thenyoumustwantthepoortopayatleastthesamepenaltyastherichforbeing
childless.
We can recognize System 1 at work. It delivers an immediate response to any
question about rich and poor: when in doubt, favor the poor. The surprising aspect of
Schelling’s problem is that this apparently simple moral rule does not work reliably. It
generatescontradictoryanswerstothesameproblem,dependingonhowthatproblemis
framed.Andofcourseyoualreadyknowthequestionthatcomesnext.Nowthatyouhave
seenthatyourreactionstotheproblemareinfluencedbytheframe,whatisyouranswerto
thequestion:Howshouldthetaxcodetreatthechildrenoftherichandthepoor?
Hereagain,youwillprobablyfindyourselfdumbfounded.Youhavemoralintuitions
aboutdifferencesbetweentherichandthepoor,buttheseintuitionsdependonanarbitrary
reference point, and they are not about the real problem. This problem—the question
aboutactualstatesoftheworld—ishowmuchtaxindividualfamiliesshouldpay,howto
fill the cells in the matrix of the tax code. You have no compelling moral intuitions to
guide you in solving that problem. Your moral feelings are attached to frames, to
descriptionsofrealityratherthantorealityitself.Themessageaboutthenatureofframing
is stark: framing should not be viewed as an intervention that masks or distorts an
underlying preference. At least in this instance—and also in the problems of the Asian
diseaseandofsurgeryversusradiationforlungcancer—thereisnounderlyingpreference
thatismaskedordistortedbytheframe.Ourpreferencesareaboutframedproblems,and
ourmoralintuitionsareaboutdescriptions,notaboutsubstance.
GoodFrames
Not all frames are equal, and s Bon nd t=“4%” wome frames are clearly better than
alternativewaystodescribe(ortothinkabout)thesamething.Considerthefollowingpair
ofproblems:
Awomanhasboughttwo$80ticketstothetheater.Whenshearrivesatthetheater,
she opens her wallet and discovers that the tickets are missing. Will she buy two
moreticketstoseetheplay?
Awoman goes tothe theater, intendingtobuy twoticketsthat cost$80 each. She
arrivesatthetheater,opensherwallet,anddiscoverstoherdismaythatthe$160with
whichshewasgoingtomakethepurchaseismissing.Shecouldusehercreditcard.
Willshebuythetickets?
Respondents who see only one version of this problem reach different conclusions,
depending on the frame. Most believe that the woman in the first story will go home
without seeing the show if she has lost tickets, and most believe that she will charge
ticketsfortheshowifshehaslostmoney.
Theexplanationshouldalreadybefamiliar—thisprobleminvolvesmentalaccounting
andthesunk-costfallacy.Thedifferentframesevokedifferentmentalaccounts,andthe
significanceofthelossdependsontheaccounttowhichitisposted.Whenticketstoa
particularshowarelost,itisnaturaltopostthemtotheaccountassociatedwiththatplay.
Thecostappearstohavedoubledandmaynowbemorethantheexperienceisworth.In
contrast,alossofcashischargedtoa“generalrevenue”account—thetheaterpatronis
slightlypoorerthanshehadthoughtshewas,andthequestionsheislikelytoaskherself
is whether the small reduction in her disposable wealth will change her decision about
payingfortickets.Mostrespondentsthoughtitwouldnot.
Theversioninwhichcashwaslostleadstomorereasonabledecisions.Itisabetter
frame because the loss, even if tickets were lost, is “sunk,” and sunk costs should be
ignored. History is irrelevant and the only issue that matters is the set of options the
theaterpatronhasnow,andtheirlikelyconsequences.Whatevershelost,therelevantfact
isthatsheislesswealthythanshewasbeforesheopenedherwallet.Ifthepersonwho
lostticketsweretoaskformyadvice,thisiswhatIwouldsay:“Wouldyouhavebought
ticketsifyouhadlosttheequivalentamountofcash?Ifyes,goaheadandbuynewones.”
Broaderframesandinclusiveaccountsgenerallyleadtomorerationaldecisions.
Inthenextexample,twoalternativeframesevokedifferentmathematicalintuitions,
and one is much superior to the other. In an article titled “The MPG Illusion,” which
appearedinScience magazine in2008, thepsychologists Richard Larrickand JackSoll
identifiedacaseinwhichpassiveacceptanceofamisleadingframehassubstantialcosts
andseriouspolicyconsequences.Mostcarbuyerslistgasmileageasoneofthe factors
thatdeterminetheirchoice;theyknowthathigh-mileagecarshaveloweroperatingcosts.
Buttheframe thathastraditionally been usedin theUnitedStates—milesper gallon—
provides very poor guidance to the decisions of both individuals and policy makers.
Considertwocarownerswhoseektoreducetheircosts:
Adamswitchesfromagas-guzzlerof12mpgtoaslightlylessvoraciousguzzlerthat
runsat14mpg.
TheenvironmentallyvirtuousBethswitchesfromaBonssesfrom30mpgcartoone
thatrunsat40mpg.
Suppose both drivers travel equal distances over a year. Who will save more gas by
switching?YoualmostcertainlysharethewidespreadintuitionthatBeth’sactionismore
significantthanAdam’s:shereducedmpgby10milesratherthan2,andbyathird(from
30to40)ratherthanasixth(from12to14).NowengageyourSystem2andworkitout.
Ifthetwocarownersbothdrive10,000miles,Adamwillreducehisconsumptionfroma
scandalous833gallonstoastillshocking714gallons,forasavingof119gallons.Beth’s
useoffuelwilldropfrom333gallonsto250,savingonly83gallons.Thempgframeis
wrong, and it should be replaced by the gallons-per-mile frame (or liters-per–100
kilometers, which is used in most other countries). As Larrick and Soll point out, the
misleading intuitions fostered by the mpg frame are likely to mislead policy makers as
wellascarbuyers.
Under President Obama, Cass Sunstein served as administrator of the Office of
Information and Regulatory Affairs. With Richard Thaler, Sunstein coauthored Nudge,
whichisthebasicmanualforapplyingbehavioraleconomicstopolicy.Itwasnoaccident
thatthe“fueleconomyandenvironment”stickerthatwillbedisplayedoneverynewcar
starting in 2013 will for the first time in the United States include the gallons-per-mile
information.Unfortunately,thecorrectformulationwillbeinsmallprint,alongwiththe
morefamiliarmpginformationinlargeprint,butthemoveisintherightdirection.The
five-yearintervalbetweenthepublicationof“TheMPGIllusion”andtheimplementation
of a partial correction is probably a speed record for a significant application of
psychologicalsciencetopublicpolicy.
A directive about organ donation in case of accidental death is noted on an
individual’sdriverlicenseinmanycountries.Theformulationofthatdirectiveisanother
caseinwhichoneframeisclearlysuperiortotheother.Fewpeoplewouldarguethatthe
decision of whether or not to donate one’s organs is unimportant, but there is strong
evidencethat most peoplemake theirchoice thoughtlessly. The evidencecomes froma
comparisonoftherate oforgandonationin Europeancountries,whichreveals startling
differencesbetweenneighboringandculturallysimilarcountries.Anarticlepublishedin
2003notedthattherateoforgandonationwascloseto100%inAustriabutonly12%in
Germany,86%inSwedenbutonly4%inDenmark.
Theseenormousdifferencesareaframingeffect,whichiscausedbytheformatofthe
critical question. The high-donation countries have an opt out form, where individuals
who wish not to donate must check an appropriate box. Unless they take this simple
action,theyareconsideredwillingdonors.Thelow-contributioncountrieshaveanopt-in
form:youmustcheckaboxtobecomeadonor.Thatisall.Thebestsinglepredictorof
whetherornotpeoplewilldonatetheirorgansisthedesignationofthedefaultoptionthat
willbeadoptedwithouthavingtocheckabox.
UnlikeotherframingeffectsthathavebeentracedtofeaturesofSystem1,theorgan
donationeffectisbestexplainedbythelazinessofSystem2.Peoplewillchecktheboxif
theyhavealreadydecidedwhattheywishtodo.Iftheyareunpreparedforthequestion,
theyhavetomaketheeffortofthinkingwhethertheywanttocheckthebox.Iimaginean
organdonationforminwhichpeoplearerequiredtosolveamathematicalprobleminthe
boxthatcorrespondstotheirdecision.Oneoftheboxescontainstheproblem2+2=?
Theproblemintheotherboxis13×37=?Therateofdonationswouldsurelybeswayed.
When the role of formulation is acknowledged, a policy question arises: Which
formulationshouldbeadopted?Inthiscase,theanswerisstraightforward.Ifyoubelieve
thatalargesupplyofdonatedorgansisgoodforsociety,youwillnotbeneutralbetweena
formulation that yields almost 100% donations and another formulation that elicits
donationsfrom4%ofdrivers.
As we have seen again and again, an important choice is controlled by an utterly
inconsequential feature of the situation. This is embarrassing—it is not how we would
wishtomakeimportantdecisions.Furthermore,itisnothowweexperiencetheworkings
ofourmind,buttheevidenceforthesecognitiveillusionsisundeniable.
Countthatasapointagainsttherational-agenttheory.Atheorythatisworthyofthe
nameassertsthatcertaineventsareimpossible—theywillnothappenifthetheoryistrue.
Whenan“impossible”eventisobserved,thetheoryisfalsified.Theoriescansurvivefora
longtimeafterconclusiveevidencefalsifiesthem,andtherational-agentmodelcertainly
survivedtheevidencewehaveseen,andmuchotherevidenceaswell.
Thecaseoforgandonationshowsthatthedebateabouthumanrationalitycanhavea
largeeffectinthereal world. A significantdifferencebetweenbelieversintherational-
agentmodelandtheskepticswhoquestionitisthatthebelieverssimplytakeitforgranted
that the formulation of a choice cannot determine preferences on significant problems.
Theywillnotevenbeinterestedininvestigatingtheproblem—andso we are oftenleft
withinferioroutcomes.
Skeptics about rationality are not surprised. They are trained to be sensitive to the
powerofinconsequentialfactorsasdeterminantsofpreference—myhopeisthatreaders
ofthisbookhaveacquiredthissensitivity.
SpeakingofFramesandReality
“Theywillfeelbetteraboutwhathappenediftheymanagetoframetheoutcomein
termsofhowmuchmoneytheykeptratherthanhowmuchtheylost.”
“Let’sreframetheproblembychangingthereferencepoint.Imaginewedidnotown
it;howmuchwouldwethinkitisworth?”
“Chargethelosstoyourmentalaccountof‘generalrevenue’—youwillfeelbetter!”
“Theyaskyoutochecktheboxtooptoutoftheirmailinglist.Theirlistwouldshrink
iftheyaskedyoutocheckaboxtooptin!”
P
Part5
P
TwoSelves
P
TwoSelves
Thetermutilityhashadtwodistinctmeaningsinitslonghistory.JeremyBenthamopened
his Introduction to the Principles of Morals and Legislation with the famous sentence
“Nature has placed mankind under the governance of two sovereign masters, pain and
pleasure.Itisforthemalonetopointoutwhatweoughttodo,aswellastodetermine
whatweshalldo.”Inanawkwardfootnote,Benthamapologizedforapplyingtheword
utility to these experiences, saying that he had been unable to find a better word. To
distinguishBentham’sinterpretationoftheterm,Iwillcallitexperiencedutility.
Forthelast100years,economistshaveusedthesamewordtomeansomethingelse.
Aseconomistsanddecisiontheoristsapplytheterm,itmeans“wantability”—andIhave
calleditdecisionutility.Expectedutilitytheory,forexample,isentirelyabouttherulesof
rationalitythatshouldgoverndecisionutilities;ithasnothingatalltosayabouthedonic
experiences.Ofcourse,thetwoconceptsofutilitywillcoincideifpeoplewantwhatthey
willenjoy,andenjoywhattheychoseforthemselves—andthisassumptionofcoincidence
is implicit in the general idea that economic agents are rational. Rational agents are
expected to know their tastes, both present and future, and they are supposed to make
gooddecisionsthatwillmaximizetheseinterests.
ExperiencedUtility
Myfascinationwiththepossiblediscrepanciesbetweenexperiencedutilityanddecision
utilitygoesbackalongway.WhileAmosandIwerestillworkingonprospecttheory,I
formulatedapuzzle,whichwentlikethis:imagineanindividualwhoreceivesonepainful
injectioneveryday.Thereisnoadaptation;thepainisthesamedaytoday.Willpeople
attachthesamevaluetoreducingthenumberofplannedinjectionsfrom20to18asfrom
6to4?Isthereanyjustificationforadistinction?
Ididnotcollectdata,becausetheoutcomewasevident.Youcanverifyforyourself
thatyouwouldpaymoretoreducethenumberofinjectionsbyathird(from6to4)than
byonetenth(from20to18).Thedecisionutilityofavoidingtwoinjectionsishigherin
thefirstcasethaninthesecond,andeveryonewillpaymoreforthefirstreductionthan
forthesecond.Butthisdifferenceisabsurd.Ifthepaindoesnotchangefromdaytoday,
whatcouldjustifyassigningdifferentutilitiestoareductionofthetotalamountofpainby
twoinjections,dependingonthenumberofpreviousinjections?Inthetermswewould
usetoday,thepuzzleintroducedtheideathatexperiencedutilitycouldbemeasuredbythe
numberofinjections.Italsosuggestedthat,atleastinsomecases,experiencedutilityis
thecriterionbywhichadecisionshouldbeassessed.Adecisionmakerwhopaysdifferent
amountstoachievethesamegainofexperiencedutility(orbesparedthesameloss)is
makingamistake.Youmayfindthisobservationobvious,butindecisiontheorytheonly
basisforjudgingthatadecisioniswrongisinconsistencywithotherpreferences.Amos
andIdiscussedtheproblembutwedidnotpursueit.Manyyearslater,Ireturnedtoit.
ExperienceandMemory
How can experienced utility be measured? How should we answer questions such as
“How much pain did Helen suffer during the medical procedure?” or “How much
enjoymentdidshegetfromher20minutesonthebeach?”TJonet8221;TJheBritish
economist Francis Edgeworth speculated about this topic in the nineteenth century and
proposedtheideaofa“hedonimeter,”animaginaryinstrumentanalogoustothedevices
usedinweather-recordingstations,whichwouldmeasurethelevelofpleasureorpainthat
anindividualexperiencesatanymoment.
Experiencedutilitywouldvary,muchasdailytemperatureorbarometricpressuredo,
andtheresultswouldbeplottedasafunctionoftime.Theanswertothequestionofhow
muchpainorpleasureHelenexperiencedduringhermedicalprocedureorvacationwould
be the “area under the curve.” Time plays a critical role in Edgeworth’s conception. If
Helen stays on the beach for 40 minutes instead of 20, and her enjoyment remains as
intense, then the total experienced utility of that episode doubles, just as doubling the
number of injections makes a course of injections twice as bad. This was Edgeworth’s
theory,andwenowhaveapreciseunderstandingoftheconditionsunderwhichhistheory
holds.
Thegraphsinfigure15showprofilesoftheexperiencesoftwopatientsundergoinga
painful colonoscopy, drawn from a study that Don Redelmeier and I designed together.
Redelmeier,aphysicianandresearcherattheUniversityofToronto,carrieditoutinthe
early1990s.Thisprocedureisnowroutinelyadministeredwithananestheticaswellasan
amnesicdrug,butthesedrugswerenotaswidespreadwhenourdatawerecollected.The
patientswerepromptedevery60secondstoindicatethelevelofpaintheyexperiencedat
the moment. The data shown are on a scale where zero is “no pain at all” and 10 is
“intolerable pain.” As you can see, the experience of each patient varied considerably
duringtheprocedure,whichlasted8minutesforpatientAand24minutesforpatientB
(thelastreadingofzeropainwasrecordedaftertheendoftheprocedure).Atotalof154
patientsparticipatedintheexperiment;theshortestprocedurelasted4minutes,thelongest
69minutes.
Next,consideraneasyquestion:Assumingthatthetwopatientsusedthescaleofpain
similarly,whichpatientsufferedmore?Nocontest.Thereisgeneralagreementthatpatient
Bhadtheworsetime.PatientBspentatleastasmuchtimeaspatientAatanylevelof
pain,andthe“areaunderthecurve”isclearlylargerforBthanforA.Thekeyfactor,of
course,isthatB’sprocedurelastedmuchlonger.Iwillcallthemeasuresbasedonreports
ofmomentarypainhedonimetertotals.
Figure15
Whentheprocedurewasover,allparticipantswereaskedtorate“thetotalamountof
pain”theyhadexperiencedduringtheprocedure.Thewordingwasintendedtoencourage
themtothinkoftheintegralofthepaintheyhadreported,reproducingthehedonimeter
totals.Surprisingly,thepatientsdidnothingofthekind.Thestatisticalanalysisrevealed
twofindings,whichillustrateapatternwehaveobservedinotherexperiments:
Peak-endrule:Theglobalretrospectiveratingwaswellpredictedbytheaverageof
thelevelofpainreportedattheworstmomentoftheexperienceandatitsend.
Duration neglect: The duration of the procedure had no effect whatsoever on the
ratingsoftotalpain.
YoucannowapplytheserulestotheprofilesofpatientsAandB.TheworstratiJoner
soeveronng(8onthe10-pointscale)wasthesameforbothpatients,butthelastrating
beforetheendoftheprocedurewas7forpatientAandonly1forpatientB.Thepeak-end
averagewastherefore7.5forpatientAandonly4.5forpatientB.Asexpected,patientA
retained a much worse memory of the episode than patient B. It was the bad luck of
patient A that the procedure ended at a bad moment, leaving him with an unpleasant
memory.
Wenowhaveanembarrassmentofriches:twomeasuresofexperiencedutility—the
hedonimetertotalandtheretrospectiveassessment—thataresystematicallydifferent.The
hedonimeter totals are computed by an observer from an individual’s report of the
experience of moments. We call these judgments duration-weighted, because the
computation of the “area under the curve” assigns equal weights to all moments: two
minutes of pain at level 9 is twice as bad as one minute at the same level of pain.
However, the findings of this experiment and others show that the retrospective
assessmentsareinsensitivetodurationandweighttwosingularmoments,thepeakandthe
end,muchmorethanothers.Sowhichshouldmatter?Whatshouldthephysiciando?The
choicehasimplicationsformedicalpractice.Wenotedthat:
Iftheobjectiveistoreducepatients’memoryofpain,loweringthepeakintensityof
paincouldbemoreimportantthanminimizingthedurationoftheprocedure.Bythe
samereasoning,gradualreliefmaybepreferabletoabruptreliefifpatientsretaina
bettermemorywhenthepainattheendoftheprocedureisrelativelymild.
Iftheobjectiveistoreducetheamountofpainactuallyexperienced,conductingthe
procedure swiftly may be appropriate even if doing so increases the peak pain
intensityandleavespatientswithanawfulmemory.
Whichofthetwoobjectivesdidyoufindmostcompelling?Ihavenotconductedaproper
survey,butmyimpressionisthatastrongmajoritywillcomedowninfavorofreducing
the memory of pain. I find it helpful to think of this dilemma as a conflict of interests
between two selves (which do not correspond to the two familiar systems). The
experiencing self is the one that answers the question: “Does it hurt now?” The
remembering self is the one that answers the question: “How was it, on the whole?”
Memoriesareallwegettokeepfromourexperienceofliving,andtheonlyperspective
thatwecanadoptaswethinkaboutourlivesisthereforethatoftherememberingself.
A comment I heard from a member of the audience after a lecture illustrates the
difficulty of distinguishing memories from experiences. He told of listening raptly to a
longsymphonyon a discthatwasscratchedneartheend,producingashockingsound,
andhereportedthatthebadending“ruinedthewholeexperience.”Buttheexperiencewas
notactuallyruined,onlythememoryofit.Theexperiencingselfhadhadanexperience
thatwasalmostentirelygood,andthebadendcouldnotundoit,becauseithadalready
happened.Myquestionerhadassigned the entireepisodeafailinggradebecause it had
endedverybadly,butthatgradeeffectivelyignored40minutesofmusicalbliss.Doesthe
actualexperiencecountfornothing?
Confusingexperiencewiththememoryofitisacompellingcognitiveillusion—and
it is the substitution that makes us believe a past experience can be ruined. The
experiencingselfdoesnothaveavoice.Therememberingselfissometimeswrong,butit
istheonethatkeepsscoreandgovernswhatwelearnfromliving,anditistheonethat
makesdecisionsJonthaperienci.Whatwelearnfromthepastistomaximizethequalities
ofourfuturememories,notnecessarilyofourfutureexperience.Thisisthetyrannyofthe
rememberingself.
WhichSelfShouldCount?
Todemonstratethedecision-makingpoweroftherememberingself,mycolleaguesandI
designed an experiment, using a mild form of torture that I will call the cold-hand
situation(itsuglytechnicalnameiscold-pressor).Participantsareaskedtoholdtheirhand
uptothewristinpainfullycoldwateruntiltheyareinvitedtoremoveitandareoffereda
warmtowel.Thesubjectsinourexperimentusedtheirfreehandtocontrolarrowsona
keyboard to provide a continuous record of the pain they were enduring, a direct
communicationfromtheirexperiencingself.Wechoseatemperaturethatcausedmoderate
buttolerablepain:thevolunteerparticipantswereofcoursefreetoremovetheirhandat
anytime,butnonechosetodoso.
Eachparticipantenduredtwocold-handepisodes:
The short episode consisted of 60 seconds of immersion in water at 14° Celsius,
which is experienced as painfully cold, but not intolerable. At the end of the 60
seconds, the experimenter instructed the participant to remove his hand from the
waterandofferedawarmtowel.
Thelongepisodelasted90seconds.Itsfirst60secondswereidenticaltotheshort
episode.Theexperimentersaidnothingatallattheendofthe60seconds.Insteadhe
openedavalvethatallowedslightlywarmerwatertoflowintothetub.Duringthe
additional30seconds,thetemperatureofthewaterrosebyroughly1°,justenough
formostsubjectstodetectaslightdecreaseintheintensityofpain.
Our participants were told that they would have three cold-hand trials, but in fact they
experiencedonlytheshortandthelongepisodes,each with a different hand.Thetrials
were separated by seven minutes. Seven minutes after the second trial, the participants
were given a choice about the third trial. They were told that one of their experiences
wouldberepeatedexactly,andwerefreetochoosewhethertorepeattheexperiencethey
hadhadwiththeirlefthandorwiththeirrighthand.Ofcourse,halftheparticipantshad
theshorttrialwiththelefthand,halfwiththeright;halfhadtheshorttrialfirst,halfbegan
withthelong,etc.Thiswasacarefullycontrolledexperiment.
The experiment was designed to create a conflict between the interests of the
experiencing and the remembering selves, and also between experienced utility and
decisionutility.Fromtheperspectiveoftheexperiencingself,thelongtrialwasobviously
worse. We expected the remembering self to have another opinion. The peak-end rule
predictsaworsememoryfortheshortthanforthelongtrial,anddurationneglectpredicts
that the difference between 90 seconds and 60 seconds of pain will be ignored. We
thereforepredictedthattheparticipantswouldhaveamorefavorable(orlessunfavorable)
memoryofthelongtrialandchoosetorepeatit.Theydid.Fully80%oftheparticipants
whoreportedthattheirpaindiminishedduringthefinalphaseofthelongerepisodeopted
torepeatit,therebydeclaringthemselveswillingtosuffer30secondsofneedlesspainin
theanticipatedthirdtrial.
The subjects who preferred the long episode were not masochists and did not
deliberately choose to expose themselves to the worse experience; they simply Jon the
heigmadeamistake.Ifwehadaskedthem,“Wouldyouprefera90-secondimmersionor
onlythefirstpartofit?”theywouldcertainlyhaveselectedtheshortoption.Wedidnot
usethesewords,however,andthesubjectsdidwhatcamenaturally:theychosetorepeat
the episode of which they had the less aversive memory. The subjects knew quite well
which of the two exposures was longer—we asked them—but they did not use that
knowledge. Their decision was governed by a simple rule of intuitive choice: pick the
optionyoulikethemost,ordisliketheleast.Rulesofmemorydeterminedhowmuchthey
disliked the two options, which in turn determined their choice. The cold-hand
experiment,likemyoldinjectionspuzzle,revealedadiscrepancybetweendecisionutility
andexperiencedutility.
Thepreferencesweobservedinthisexperimentareanotherexampleoftheless-is-
moreeffectthatwehaveencounteredonpreviousoccasions.OnewasChristopherHsee’s
studyinwhichaddingdishestoasetof24dishesloweredthetotalvaluebecausesomeof
theaddeddisheswerebroken.AnotherwasLinda,theactivistwomanwhoisjudgedmore
likelytobeafeministbanktellerthanabankteller.Thesimilarityisnotaccidental.The
sameoperatingfeatureofSystem1accountsforallthreesituations:System1represents
setsbyaverages,norms,andprototypes,notbysums.Eachcold-handepisodeisasetof
moments, which the remembering self stores as a prototypical moment. This leads to a
conflict. For an objective observer evaluating the episode from the reports of the
experiencingself,whatcountsisthe“areaunderthecurve”thatintegratespainovertime;
ithasthenatureofasum.Thememorythattherememberingselfkeeps,incontrast,isa
representativemoment,stronglyinfluencedbythepeakandtheend.
Of course, evolution could have designed animals’ memory to store integrals, as it
surelydoesinsomecases.Itisimportantforasquirrelto“know”thetotalamountoffood
it has stored, and a representation of the average size of the nuts would not be a good
substitute.However, the integral of pain or pleasure overtime maybe lessbiologically
significant.Weknow,forexample,thatratsshowdurationneglectforbothpleasureand
pain.Inoneexperiment,ratswereconsistentlyexposedtoasequenceinwhichtheonset
ofalightsignalsthatanelectricshockwillsoonbedelivered.Theratsquicklylearnedto
fearthelight,andtheintensityoftheirfearcouldbemeasuredbyseveralphysiological
responses.Themainfindingwasthatthedurationoftheshockhaslittleornoeffecton
fear—allthatmattersisthepainfulintensityofthestimulus.
Other classic studies showed that electrical stimulation of specific areas in the rat
brain (and of corresponding areas in the human brain) produce a sensation of intense
pleasure,so intensein somecases thatrats whocan stimulatetheir brainby pressinga
leverwilldieofstarvationwithouttakingabreaktofeedthemselves.Pleasurableelectric
stimulationcanbedeliveredinburststhatvaryinintensityandduration.Hereagain,only
intensitymatters.Uptoapoint,increasingthedurationofaburstofstimulationdoesnot
appear to increase the eagerness of the animal to obtain it. The rules that govern the
rememberingselfofhumanshavealongevolutionaryhistory.
Biologyvs.Rationality
Themostusefulideaintheinjectionspuzzlethatpreoccupiedmeyearsagowasthatthe
experienced utility of a series of equally painful injections can be measured, by simply
countingtheinjections.Ifallinjectionsareequallyaversive,then20ofthemaretwiceas
badas10,andJoneoeeareductionfrom20to18andareductionfrom6to4areequally
valuable. If the decision utility does not correspond to the experienced utility, then
something is wrong with the decision. The same logic played out in the cold-hand
experiment:anepisodeofpainthatlasts90secondsisworsethanthefirst60secondsof
thatepisode.Ifpeoplewillinglychoosetoendurethelongerepisode,somethingiswrong
with their decision. In my early puzzle, the discrepancy between the decision and the
experienceoriginated fromdiminishing sensitivity:the differencebetween 18and 20is
less impressive, and appears to be worth less, than the difference between 6 and 4
injections. In the cold-hand experiment, the error reflects two principles of memory:
durationneglectandthepeak-endrule.Themechanismsaredifferentbuttheoutcomeis
thesame:adecisionthatisnotcorrectlyattunedtotheexperience.
Decisionsthatdonotproducethebestpossibleexperienceanderroneousforecastsof
future feelings—both are bad news for believers in the rationality of choice. The cold-
handstudyshowedthatwecannotfullytrustourpreferencestoreflectourinterests,even
iftheyarebasedonpersonalexperience,andevenifthememoryofthatexperiencewas
laiddownwithinthelastquarterofanhour!Tastesanddecisionsareshapedbymemories,
andthememoriescanbewrong.Theevidencepresentsaprofoundchallengetotheidea
thathumanshaveconsistentpreferencesandknowhowtomaximizethem,acornerstone
ofthe rational-agentmodel.An inconsistencyis builtintothe designof ourminds. We
havestrongpreferencesaboutthe duration of ourexperiencesofpainand pleasure.We
wantpaintobebriefandpleasuretolast.Butourmemory,afunctionofSystem1,has
evolvedtorepresentthemostintensemomentofanepisodeofpainorpleasure(thepeak)
andthefeelingswhentheepisodewasatitsend.Amemorythatneglectsdurationwillnot
serveourpreferenceforlongpleasureandshortpains.
SpeakingofTwoSelves
“You are thinking of your failed marriage entirely from the perspective of the
rememberingself.Adivorceislikeasymphonywithascreechingsoundattheend—
thefactthatitendedbadlydoesnotmeanitwasallbad.”
“Thisisabadcaseofdurationneglect.Youaregivingthegoodandthebadpartof
yourexperienceequalweight,althoughthegoodpartlastedtentimesaslongasthe
other.”
P
LifeasaStory
Earlyinthedaysofmyworkonthemeasurementofexperience,IsawVerdi’soperaLa
Traviata.Knownforitsgorgeousmusic,itisalsoamovingstoryofthelovebetweena
young aristocrat and Violetta, a woman of the demimonde. The young man’s father
approaches Violetta and convinces her to give up her lover, to protect the honor of the
familyandthemarriageprospects of theyoungman’ssister. In anactofsupreme self-
sacrifice, Violetta pretends to reject the man she adores. She soon relapses into
consumption(thenineteenth-centurytermfortuberculosis).Inthefinalact,Violettalies
dying,surroundedbyafewfriends.HerbelovedhasbeenalertedandisrushingtoParisto
see her. H Kto earing the news, she is transformed with hope and joy, but she is also
deterioratingquickly.
Nomatterhowmanytimesyouhaveseentheopera,youaregrippedbythetension
andfearof themoment:Willtheyoungloverarriveintime?Thereisasensethatitis
immenselyimportantforhimtojoinhisbelovedbeforeshedies.Hedoes,ofcourse,some
marvelousloveduetsaresung,andafter10minutesofgloriousmusicViolettadies.
Onmywayhomefromtheopera,Iwondered:Whydowecaresomuchaboutthose
last10minutes?IquicklyrealizedthatIdidnotcareatallaboutthelengthofVioletta’s
life.IfIhadbeentoldthatshediedatage27,notage28asIbelieved,thenewsthatshe
hadmissedayearofhappylifewouldnot have movedmeatall,butthepossibilityof
missingthelast10minutesmatteredagreatdeal.Furthermore,theemotionIfeltaboutthe
lovers’ reunion would not have changed if I had learned that they actually had a week
together, rather than 10 minutes. If the lover had come too late, however, La Traviata
would have been an altogether different story. A story is about significant events and
memorablemoments,notabouttimepassing.Durationneglectisnormalinastory,and
the ending often defines its character. The same core features appear in the rules of
narratives and in the memories of colonoscopies, vacations, and films. This is how the
rememberingselfworks:itcomposesstoriesandkeepsthemforfuturereference.
Itisnotonlyattheoperathatwethinkoflifeasastoryandwishittoendwell.When
wehearaboutthedeathofawomanwhohadbeenestrangedfromherdaughterformany
years, we want to know whether they were reconciled as death approached. We do not
care only about the daughter’s feelings—it is the narrative of the mother’s life that we
wishtoimprove.Caringforpeopleoftentakestheformofconcernforthequalityoftheir
stories,notfortheirfeelings.Indeed,wecanbedeeplymovedevenbyeventsthatchange
thestoriesofpeoplewhoarealreadydead.Wefeelpityforamanwhodiedbelievingin
hiswife’sloveforhim,whenwehearthatshehadaloverformanyyearsandstayedwith
herhusbandonlyforhismoney.Wepitythehusbandalthoughhehadlivedahappylife.
Wefeelthehumiliationofascientistwhomadeanimportantdiscoverythatwasproved
falseaftershedied,althoughshedidnotexperiencethehumiliation.Mostimportant,of
course,weallcareintenselyforthenarrativeofourownlifeandverymuchwantittobea
goodstory,withadecenthero.
ThepsychologistEdDienerandhisstudentswonderedwhetherdurationneglectand
thepeak-endrulewouldgovernevaluationsofentirelives.Theyusedashortdescription
ofthelifeofafictitiouscharactercalledJen,a never-marriedwoman with no children,
whodiedinstantlyandpainlesslyinanautomobileaccident.InoneversionofJen’sstory,
shewasextremelyhappythroughoutherlife(whichlastedeither30or60years),enjoying
herwork,takingvacations,spendingtimewithherfriendsandonherhobbies.Another
versionadded5extrayearstoJen’slife,whonowdiedeitherwhenshewas35or65.The
extrayearsweredescribedaspleasantbutlesssothanbefore.Afterreadingaschematic
biographyofJen,eachparticipantansweredtwoquestions:“Takingherlifeasawhole,
how desirable do you think Jen’s life was?” and “How much total happiness or
unhappinesswouldyousaythatJenexperiencedinherlife?”
Theresultsprovidedclearevidenceofbothdurationneglectandapeak-endeffect.In
abetween-subjects experiment(differentparticipantssaw differentforms), doubling the
durationofJen’slifehadJtoAadJtonoeffectwhatsoeveronthedesirabilityofherlife,or
onjudgmentsofthetotalhappinessthatJenexperienced.Clearly,herlifewasrepresented
byaprototypicalsliceoftime,not as asequenceoftimeslices.As a consequence,her
“total happiness” was the happiness of a typical period in her lifetime, not the sum (or
integral)ofhappinessoverthedurationofherlife.
Asexpectedfromthisidea,Dienerandhisstudentsalsofoundaless-is-moreeffect,a
strong indication that an average (prototype) has been substituted for a sum. Adding 5
“slightlyhappy”yearstoaveryhappylifecausedasubstantialdropinevaluationsofthe
totalhappinessofthatlife.
Atmyurging,theyalsocollecteddataontheeffectoftheextra5yearsinawithin-
subject experiment; each participant made both judgments in immediate succession. In
spiteofmylongexperiencewithjudgmenterrors,Ididnotbelievethatreasonablepeople
couldsaythatadding5slightlyhappyyearstoalifewouldmakeitsubstantiallyworse.I
waswrong.Theintuitionthatthedisappointingextra5yearsmadethewholelifeworse
wasoverwhelming.
The pattern of judgments seemed so absurd that Diener and his students initially
thought that it represented the folly of the young people who participated in their
experiments.However,thepatterndidnotchangewhentheparentsandolderfriendsof
students answered the same questions. In intuitive evaluation of entire lives as well as
briefepisodes,peaksandendsmatterbutdurationdoesnot.
Thepainsoflaborandthebenefitsofvacationsalwayscomeupasobjectionstothe
ideaofdurationneglect:weallsharetheintuitionthatitismuchworseforlabortolast24
than6hours,andthat6daysatagoodresortisbetterthan3.Durationappearstomatterin
thesesituations,butthisisonlybecausethequalityoftheendchangeswiththelengthof
theepisode.Themotherismoredepletedandhelplessafter24hoursthanafter6,andthe
vacationerismorerefreshedandrestedafter6daysthanafter3.Whattrulymatterswhen
weintuitivelyassesssuchepisodesistheprogressivedeteriorationorimprovementofthe
ongoingexperience,andhowthepersonfeelsattheend.
AmnesicVacations
Considerthechoiceofavacation.Doyouprefertoenjoyarelaxingweekatthefamiliar
beachto whichyou went lastyear? Ordo youhope toenrich your storeof memories?
Distinctindustrieshavedevelopedtocatertothesealternatives:resortsofferrestorative
relaxation; tourism is about helping people construct stories and collect memories. The
frenetic picture taking of many tourists suggests that storing memories is often an
importantgoal,whichshapesboththeplansforthevacationandtheexperienceofit.The
photographerdoesnotviewthesceneasamomenttobesavoredbutasafuturememory
tobedesigned.Picturesmaybeusefultotherememberingself—thoughwerarelylookat
themforverylong,orasoftenasweexpected,orevenatall—butpicturetakingisnot
necessarilythebestwayforthetourist’sexperiencingselftoenjoyaview.
Inmanycasesweevaluatetouristicvacationsbythestoryandthememoriesthatwe
expect to store. The word memorable is often used to describe vacation highlights,
explicitlyrevealingthegoaloftheexperience.Inothersituations—lovecomestomind—
the declaration that the present moment will never be forgotten, though not always
accurate,changesthecharacterofthemoment.Aself-consciouslymemorableexperience
gainsaweightandasignificanceJtoAceJtothatitwouldnototherwisehave.
EdDienerandhisteamprovidedevidencethatitistherememberingselfthatchooses
vacations.Theyaskedstudentstomaintaindailydiariesandrecordadailyevaluationof
theirexperiencesduring springbreak.The studentsalsoprovided aglobalrating ofthe
vacationwhenithadended.Finally,theyindicatedwhetherornottheyintendedtorepeat
or not to repeat the vacation they had just had. Statistical analysis established that the
intentions for future vacations were entirely determined by the final evaluation—even
when that score did not accurately represent the quality of the experience that was
describedinthediaries.Asinthecold-handexperiment,rightorwrong,peoplechooseby
memorywhentheydecidewhetherornottorepeatanexperience.
A thought experiment about your next vacation will allow you to observe your
attitudetoyourexperiencingself.
Attheend ofthe vacation, allpicturesand videoswillbe destroyed.Furthermore,
youwillswallowapotionthatwillwipeoutallyourmemoriesofthevacation.
How would this prospect affect your vacation plans? How much would you be
willingtopayforit,relativetoanormallymemorablevacation?
While I have not formally studied the reactions to this scenario, my impression from
discussingitwithpeopleisthattheeliminationofmemoriesgreatlyreducesthevalueof
the experience. In some cases, people treat themselves as they would treat another
amnesic,choosingtomaximizeoverallpleasurebyreturningtoaplacewheretheyhave
beenhappyinthepast.However,somepeoplesaythattheywouldnotbothertogoatall,
revealing that they care only about their remembering self, and care less about their
amnesicexperiencingselfthanaboutanamnesicstranger.Manypointoutthattheywould
not send either themselves or another amnesic to climb mountains or trek through the
jungle—becausetheseexperiencesaremostlypainfulinrealtimeandgainvaluefromthe
expectationthatboththepainandthejoyofreachingthegoalwillbememorable.
Foranotherthoughtexperiment,imagineyoufaceapainfuloperationduringwhich
youwillremainconscious.Youaretoldyouwillscreaminpainandbegthesurgeonto
stop.However,youarepromisedanamnesia-inducingdrugthatwillcompletelywipeout
any memory of the episode. How do you feel about such a prospect? Here again, my
informalobservationisthatmostpeopleareremarkablyindifferenttothepainsoftheir
experiencingself.Somesaytheydon’tcareatall.Otherssharemyfeeling,whichisthatI
feelpityformysufferingselfbutnotmorethanIwouldfeelforastrangerinpain.Oddas
itmayseem,Iammyrememberingself,andtheexperiencingself,whodoesmyliving,is
likeastrangertome.
SpeakingofLifeasaStory
“He is desperately trying to protect the narrative of a life of integrity, which is
endangeredbythelatestepisode.”
“Thelengthtowhichhewaswillingtogoforaone-nightencounterisasignoftotal
durationneglect.”
“You seem to be devoting your entire vacation to the construction of memories.
Perhapsyoushouldputawaythecameraandenjoythemoment,evenifitisnotvery
memorable?”
“SheisanAlzheimer’spatient.Shenolongermaintainsanarrativeofherlife,but
herexperiencingselfisstillsensitivetobeautyandgentleness.”
P
ExperiencedWell-Being
When I became interested in the study of well-being about fifteen years ago, I quickly
foundoutthatalmosteverythingthatwasknownaboutthesubjectdrewontheanswersof
millionsofpeopletominorvariationsofasurveyquestion,whichwasgenerallyaccepted
asameasureofhappiness. The questionisclearlyaddressedtoyour rememberingself,
whichisinvitedtothinkaboutyourlife:
Allthingsconsidered,howsatisfiedareyouwithyourlifeasawholethesedays?
Havingcometothetopicofwell-beingfromthestudyofthemistakenmemoriesof
colonoscopiesandpainfullycoldhands,Iwasnaturallysuspiciousofglobalsatisfaction
withlifeasavalidmeasureofwell-being.Astherememberingselfhadnotprovedtobea
goodwitnessinmyexperiments,Ifocusedonthewell-beingoftheexperiencingself.I
proposedthatitmadesensetosaythat“HelenwashappyinthemonthofMarch”if
shespentmostofhertimeengagedinactivitiesthatshewouldrathercontinuethan
stop,littletimeinsituationsshewishedtoescape,and—veryimportantbecauselife
isshort—nottoomuchtimeinaneutralstateinwhichshewouldnotcareeitherway.
Therearemanydifferentexperienceswewouldrathercontinuethanstop,including
bothmentalandphysicalpleasures.OneoftheexamplesIhadinmindforasituationthat
Helenwouldwishtocontinueistotalabsorptioninatask,whichMihalyCsikszentmihalyi
callsflow—astatethatsomeartistsexperienceintheircreativemomentsandthatmany
other people achieve when enthralled by a film, a book, or a crossword puzzle:
interruptionsarenotwelcomeinanyofthesesituations.Ialsohadmemoriesofahappy
earlychildhoodinwhichIalwayscriedwhenmymothercametotearmeawayfrommy
toystotakemetothepark,andcriedagainwhenshetookmeawayfromtheswingsand
theslide.TheresistancetointerruptionwasasignIhadbeenhavingagoodtime,both
withmytoysandwiththeswings.
I proposed to measure Helen’s objective happiness precisely as we assessed the
experienceofthetwocolonoscopypatients,byevaluatingaprofileofthewell-beingshe
experienced over successive moments of her life. In this I was following Edgeworth’s
hedonimetermethodofacenturyearlier.Inmyinitialenthusiasmforthisapproach,Iwas
inclinedtodismissHelen’srememberingselfasanerror-pronewitnesstotheactualwell-
beingofherexperiencingself.Isuspectedthispositionwastooextreme,whichitturned
outtobe,butitwasagoodstart.
n=“4”>ExperiencedWell-Being
Iassembled“adreamteam”thatincludedthreeotherpsychologistsofdifferentspecialties
andoneeconomist,andwesetouttogethertodevelopameasureofthewell-beingofthe
experiencing self. A continuous record of experience was unfortunately impossible—a
person cannot live normally while constantly reporting her experiences. The closest
alternative was experience sampling, a method that Csikszentmihalyi had invented.
Technologyhasadvancedsinceitsfirstuses.Experiencesamplingisnowimplementedby
programminganindividual’scellphonetobeeporvibrateatrandomintervalsduringthe
day. The phonethenpresents abriefmenuof questionsaboutwhatthe respondentwas
doing and who was with her when she was interrupted. The participant is also shown
ratingscalestoreporttheintensityofvariousfeelings:happiness,tension,anger,worry,
engagement,physicalpain,andothers.
Experience sampling is expensive and burdensome (although less disturbing than
most people initially expect; answering the questions takes very little time). A more
practical alternative was needed, so we developed a method that we called the Day
ReconstructionMethod(DRM).Wehopeditwouldapproximatetheresultsofexperience
sampling and provide additional information about the way people spend their time.
Participants(allwomen,intheearlystudies)wereinvitedtoatwo-hoursession.Wefirst
askedthemtorelivethepreviousdayindetail,breakingitupintoepisodeslikescenesina
film. Later, they answered menus of questions about each episode, based on the
experience-samplingmethod.Theyselectedactivitiesinwhichtheywereengagedfroma
list and indicated the one to which they paid most attention. They also listed the
individualstheyhadbeenwith,andratedtheintensityofseveralfeelingsonseparate0–6
scales (0 = the absence of the feeling; 6 = most intense feeling). Our method drew on
evidence that people who are able to retrieve a past situation in detail are also able to
relive the feelings that accompanied it, even experiencing their earlier physiological
indicationsofemotion.
We assumed that our participants would fairly accurately recover the feeling of a
prototypical moment of the episode. Several comparisons with experience sampling
confirmed the validity of the DRM. Because the participants also reported the times at
whichepisodesbeganandended,wewereabletocomputeaduration-weightedmeasure
of their feeling during the entire waking day. Longer episodes counted more than short
episodes in our summary measure of daily affect. Our questionnaire also included
measuresoflifesatisfaction,whichweinterpretedasthesatisfactionoftheremembering
self.WeusedtheDRMtostudythedeterminantsofbothemotionalwell-beingandlife
satisfactioninseveralthousandwomenintheUnitedStates,France,andDenmark.
The experience of a moment or an episode is not easily represented by a single
happiness value. There are many variants of positive feelings, including love, joy,
engagement,hope,amusement,andmanyothers.Negativeemotionsalsocomeinmany
varieties, including anger, shame, depression, and loneliness. Although positive and
negativeemotionsexistatthesametime,itispossibletoclassifymostmomentsoflifeas
ultimatelypositiveornegative.Wecouldidentifyunpleasantepisodesbycomparingthe
ratingsofpositiveandnegativeadjectives.Wecalledanepisodeunpleasantifanegative
feelingwasassignedahigherratingthanallthepositivefeelings.WefoundthatAmerican
womenspentabout19%ofthetimeinanunpleasantstate,somewhathigherthanFrench
women(16%)orDanishwomen(14%).
We called the percentage Jr”>n Qge Jr”>of time that an individual spends in an
unpleasantstatetheU-index.Forexample,anindividualwhospent4hoursofa16-hour
wakingdayinanunpleasantstatewouldhaveaU-indexof25%.TheappealoftheU-
indexisthatitisbasednotonaratingscalebutonanobjectivemeasurementoftime.If
theU-indexforapopulationdropsfrom20%to18%,youcaninferthatthetotaltimethat
thepopulationspentinemotionaldiscomfortorpainhasdiminishedbyatenth.
A striking observation was the extent of inequality in the distribution of emotional
pain. About half our participants reported going through an entire day without
experiencing an unpleasant episode. On the other hand, a significant minority of the
population experienced considerable emotional distress for much of the day. It appears
that a small fraction of the population does most of the suffering—whether because of
physical or mental illness, an unhappy temperament, or the misfortunes and personal
tragediesintheirlife.
A U-index can also be computed for activities. For example, we can measure the
proportion of time that people spend in a negative emotional state while commuting,
working, or interacting with their parents, spouses, or children. For 1,000 American
womenina Midwesterncity,theU-indexwas 29%forthemorningcommute,27%for
work,24%forchildcare,18%forhousework,12%forsocializing,12%forTVwatching,
and 5% for sex. The U-index was higher by about 6% on weekdays than it was on
weekends,mostlybecauseonweekendspeoplespendlesstimeinactivitiestheydislike
anddonotsufferthetensionandstressassociatedwithwork.Thebiggestsurprisewasthe
emotionalexperienceofthetimespentwithone’schildren,whichforAmericanwomen
wasslightlylessenjoyablethandoinghousework.Herewefoundoneofthefewcontrasts
betweenFrenchandAmericanwomen:Frenchwomenspendlesstimewiththeirchildren
butenjoyitmore,perhapsbecausetheyhavemoreaccesstochildcareandspendlessof
theafternoondrivingchildrentovariousactivities.
An individual’s mood at any moment depends on her temperament and overall
happiness, but emotional well-being also fluctuates considerably over the day and the
week.Themoodofthemomentdependsprimarilyonthecurrentsituation.Moodatwork,
for example, is largely unaffected by the factors that influence general job satisfaction,
includingbenefitsandstatus.Moreimportantaresituationalfactorssuchasanopportunity
tosocializewithcoworkers,exposuretoloudnoise,timepressure(asignificantsourceof
negativeaffect),andtheimmediatepresenceofaboss(inourfirststudy,theonlything
that was worse than being alone). Attention is key. Our emotional state is largely
determinedbywhatweattendto,andwearenormallyfocusedonourcurrentactivityand
immediateenvironment.Thereareexceptions,wherethequalityofsubjectiveexperience
isdominatedbyrecurrentthoughtsratherthanbytheeventsofthemoment.Whenhappily
in love, we may feel joy even when caught in traffic, and if grieving, we may remain
depressed when watching a funny movie. In normal circumstances, however, we draw
pleasureandpainfromwhatishappeningatthemoment,ifweattendtoit.Togetpleasure
fromeating,forexample,youmustnoticethatyouaredoingit.WefoundthatFrenchand
American women spent about the same amount of time eating, but for Frenchwomen,
eatingwastwiceaslikelytobefocalasitwasforAmericanwomen.TheAmericanswere
farmorepronetocombineeatingwithotheractivities,andtheirpleasurefromeatingwas
correspondinglydiluted.
These observations have implications for both individuals and society. The use of
timeisoneoftheareasoflifeoverwhichpeoplehavesomecontrol.Fewindividualscan
willthemselvestohaJr”>nQhaJr”>veasunnierdisposition,butsomemaybeableto
arrangetheirlivestospendlessoftheirdaycommuting,andmoretimedoingthingsthey
enjoywithpeopletheylike.Thefeelingsassociatedwithdifferentactivitiessuggestthat
another way to improve experience is to switch time from passive leisure, such as TV
watching, to more active forms of leisure, including socializing and exercise. From the
socialperspective,improved transportation forthelabor force,availabilityof child care
for working women, and improved socializing opportunities for the elderly may be
relativelyefficientwaystoreducetheU-indexofsociety—evenareductionby1%would
be a significant achievement, amounting to millions of hours of avoided suffering.
Combinednationalsurveysoftimeuseandofexperiencedwell-beingcaninformsocial
policyinmultipleways.Theeconomistonourteam,AlanKrueger,tooktheleadinan
efforttointroduceelementsofthismethodintonationalstatistics.
Measuresofexperiencedwell-beingarenowroutinelyusedinlarge-scalenationalsurveys
intheUnitedStates,Canada,andEurope,andtheGallupWorldPollhasextendedthese
measurements to millions of respondents in the United States and in more than 150
countries. The polls elicit reports of the emotions experienced during the previous day,
thoughinlessdetailthantheDRM.Thegiganticsamplesallowextremelyfineanalyses,
which have confirmed the importance of situational factors, physical health, and social
contact in experienced well-being. Not surprisingly, a headache will make a person
miserable,andthesecondbestpredictorofthefeelingsofadayiswhetherapersondidor
didnothavecontactswithfriendsorrelatives.Itisonlyaslightexaggerationtosaythat
happinessistheexperienceofspendingtimewithpeopleyouloveandwholoveyou.
TheGallupdatapermitacomparisonoftwoaspectsofwell-being:
thewell-beingthatpeopleexperienceastheylivetheirlives
thejudgmenttheymakewhentheyevaluatetheirlife
Gallup’slifeevaluationismeasuredbyaquestionknownastheCantrilSelf-Anchoring
StrivingScale:
Pleaseimaginealadderwithstepsnumberedfromzeroatthebottomto10atthetop.
Thetopoftheladderrepresentsthebestpossiblelifeforyouandthebottomofthe
ladderrepresentstheworstpossiblelifeforyou.Onwhichstepoftheladderwould
yousayyoupersonallyfeelyoustandatthistime?
Some aspects of life have more effect on the evaluation of one’s life than on the
experienceofliving.Educationalattainmentisanexample.Moreeducationisassociated
withhigherevaluationofone’slife,butnotwithgreaterexperiencedwell-being.Indeed,
atleastintheUnitedStates,themoreeducatedtendtoreporthigherstress.Ontheother
hand,illhealthhasamuchstrongeradverseeffectonexperiencedwell-beingthanonlife
evaluation.Livingwithchildrenalsoimposesasignificantcostinthecurrencyofdaily
feelings—reportsofstressandangerarecommonamongparents,buttheadverseeffects
onlifeevaluationaresmaller.Religiousparticipationalsohasrelativelygreaterfavorable
impactonbothpositiveaffectandstressreductionthanonlifeevaluation.Surprisingly,
however,religionprovidesnoreductionoffeelingsofdepressionorworry.
Ananalysisofmorethan450,000responsestotheGallup-HealthwaysWell-BeiJr”>n
QBeiJr”>ng Index, a daily surveyof 1,000Americans, providesa surprisingly definite
answer to the most frequently asked question in well-being research: Can money buy
happiness? The conclusion is that being poor makes one miserable, and that being rich
mayenhanceone’slifesatisfaction,butdoesnot(onaverage)improveexperiencedwell-
being.
Severe poverty amplifies the experienced effects of other misfortunes of life. In
particular, illness is much worse for the very poor than for those who are more
comfortable.Aheadacheincreasestheproportionreportingsadnessandworryfrom19%
to38%forindividualsinthetoptwo-thirdsoftheincomedistribution.Thecorresponding
numbers for the poorest tenth are 38% and 70%—a higher baseline level and a much
largerincrease.Significantdifferencesbetweentheverypoorandothersarealsofoundfor
theeffectsofdivorceandloneliness.Furthermore,thebeneficialeffectsoftheweekendon
experiencedwell-beingaresignificantlysmallerfortheverypoorthanformosteveryone
else.
Thesatiationlevelbeyondwhichexperiencedwell-beingnolongerincreaseswasa
householdincomeofabout$75,000inhigh-costareas(itcouldbelessinareaswherethe
costoflivingislower).Theaverageincreaseofexperiencedwell-beingassociatedwith
incomesbeyondthatlevelwaspreciselyzero.Thisissurprisingbecausehigherincome
undoubtedly permits the purchase of many pleasures, including vacations in interesting
placesandoperatickets,aswellasanimprovedlivingenvironment.Whydotheseadded
pleasuresnotshowupinreportsofemotionalexperience?Aplausibleinterpretationisthat
higher income is associated with a reduced ability to enjoy the small pleasures of life.
Thereissuggestiveevidenceinfavorofthisidea:primingstudentswiththeideaofwealth
reducesthepleasuretheirfaceexpressesastheyeatabarofchocolate!
Thereisaclearcontrastbetweentheeffectsofincomeonexperiencedwell-beingand
onlifesatisfaction.Higherincomebringswithithighersatisfaction,wellbeyondthepoint
atwhichitceasestohaveanypositiveeffectonexperience.Thegeneralconclusionisas
clear for well-being as it was for colonoscopies: people’s evaluations of their lives and
theiractualexperiencemayberelated,buttheyarealsodifferent.Lifesatisfactionisnota
flawed measure of their experienced well-being, as I thought some years ago. It is
somethingelseentirely.
SpeakingofExperiencedWell-Being
“Theobjectiveofpolicyshouldbetoreducehumansuffering.WeaimforalowerU-
indexinsociety.Dealingwithdepressionandextremepovertyshouldbeapriority.”
“Theeasiestwaytoincreasehappinessistocontrolyouruseoftime.Canyoufind
moretimetodothethingsyouenjoydoing?”
“Beyond the satiation level of income, you can buy more pleasurable experiences,
butyouwilllosesomeofyourabilitytoenjoythelessexpensiveones.”
P
ThinkingAboutLife
Figure16istakenfromananalysisbyAndrewClark,EdDiener,andYannisGeorgellisof
theGermanSocio-EconomicPanel,inwhichthesamerespondentswereaskedeveryyear
abouttheirsatisfactionwiththeirlife.Respondentsalsoreportedmajorchangesthathad
occurredintheircircumstancesduringtheprecedingyear.Thegraphshowsthelevelof
satisfactionreportedbypeoplearoundthetimetheygotmarried.
Figure16
Thegraphreliablyevokesnervouslaughterfromaudiences, and the nervousnessis
easytounderstand:afterall,peoplewhodecidetogetmarrieddosoeitherbecausethey
expectitwillmakethemhappierorbecausetheyhopethatmakingatiepermanentwill
maintain the present state of bliss. In the useful term introduced by Daniel Gilbert and
TimothyWilson,thedecisiontogetmarriedreflects,formanypeople,amassiveerrorof
affectiveforecasting.Ontheirweddingday,thebrideandthegroomknowthattherateof
divorceishighandthattheincidenceofmaritaldisappointmentisevenhigher,buttheydo
notbelievethatthesestatisticsapplytothem.
Thestartlingnewsoffigure16isthesteepdeclineoflifesatisfaction.Thegraphis
commonly interpreted as tracing a process of adaptation, in which the early joys of
marriage quickly disappear as the experiences become routine. However, another
approachispossible,whichfocusesonheuristicsofjudgment.Hereweaskwhathappens
inpeople’smindswhentheyareaskedtoevaluatetheirlife.Thequestions“Howsatisfied
are you with your life as a whole?” and “How happy are you these days?” are not as
simple as “What is your telephone number?” How do survey participants manage to
answersuchquestionsinafewseconds,asalldo?Itwillhelptothinkofthisasanother
judgment.Asis also thecasefor otherquestions,somepeople may haveaready-made
answer,whichtheyhadproducedonanotheroccasioninwhichtheyevaluatedtheirlife.
Others,probablythemajority,donotquicklyfindaresponsetotheexactquestionthey
wereasked,andautomaticallymaketheirtaskeasierbysubstitutingtheanswertoanother
question. System 1 is at work. When we look at figure 16 in this light, it takes on a
differentmeaning.
Theanswerstomanysimplequestionscanbesubstitutedforaglobalevaluationof
life.Yourememberthestudyinwhichstudentswhohadjustbeenaskedhowmanydates
theyhadinthepreviousmonthreportedtheir“happinessthesedays”asifdatingwasthe
only significant fact in their life. In another well-known experiment in the same vein,
NorbertSchwarzandhiscolleaguesinvitedsubjectstothelabtocompleteaquestionnaire
onlifesatisfaction.Beforetheybeganthattask,however,heaskedthemtophotocopya
sheet of paper for him. Half the respondents found a dime on the copying machine,
planted there by the experimenter. The minor lucky incident caused a marked
improvementinsubjects’reportedsatisfactionwiththeirlifeasawhole!Amoodheuristic
isonewaytoanswerlife-satisfactionquestions.
The dating survey and the coin-on-the-machine experiment demonstrated, as
intended,thattheresponsestoglobalwell-beingquestionsshouldbetakenwithagrainof
salt.Butofcourseyourcurrentmoodisnottheonlythingthatcomestomindwhenyou
areaskedtoevaluateyourlife.Youarelikelytoberemindedofsignificanteventsinyour
recentpastornearfuture;ofrecurrentconcerns,suchasthehealthJghtA5althJghtofa
spouse or the bad company that your teenager keeps; of important achievements and
painful failures. A few ideas that are relevant to the question will occur to you; many
otherswillnot.Evenwhenitisnotinfluencedbycompletelyirrelevantaccidentssuchas
thecoinonthemachine,thescorethatyouquicklyassigntoyourlifeisdeterminedbya
smallsampleofhighlyavailableideas,notbyacarefulweightingofthedomainsofyour
life.
Peoplewhorecentlymarried,orareexpectingtomarryinthenearfuture,arelikelyto
retrieve that fact when asked a general question about their life. Because marriage is
almostalwaysvoluntaryintheUnitedStates,almosteveryonewhoisremindedofhisor
herrecentorforthcomingmarriagewillbehappywiththeidea.Attentionisthekeytothe
puzzle.Figure16canbereadasagraphofthelikelihoodthatpeoplewillthinkoftheir
recentorforthcomingmarriagewhenaskedabouttheirlife.Thesalienceofthisthoughtis
boundtodiminishwiththepassageoftime,asitsnoveltywanes.
Thefigureshows an unusuallyhighlevel of lifesatisfactionthatlasts twoorthree
years around the event of marriage. However, if this apparent surge reflects the time
courseofaheuristicforansweringthequestion,thereislittlewecanlearnfromitabout
eitherhappinessorabouttheprocessofadaptationtomarriage.Wecannotinferfromit
thatatideofraisedhappinesslastsforseveralyearsandgraduallyrecedes.Evenpeople
whoarehappytoberemindedoftheirmarriagewhenaskedaquestionabouttheirlifeare
notnecessarilyhappiertherestofthetime.Unlesstheythinkhappythoughtsabouttheir
marriage during much of their day, it will not directly influence their happiness. Even
newlywedswhoareluckyenoughtoenjoyastateofhappypreoccupationwiththeirlove
willeventuallyreturntoearth,andtheirexperiencedwell-beingwillagaindepend,asit
doesfortherestofus,ontheenvironmentandactivitiesofthepresentmoment.
In the DRM studies, there was no overall difference in experienced well-being
betweenwomenwholivedwithamateandwomenwhodidnot.Thedetailsofhowthe
twogroupsusedtheirtimeexplainedthefinding.Womenwhohaveamatespendlesstime
alone,butalsomuchlesstimewithfriends.Theyspendmoretimemakinglove,whichis
wonderful,butalsomoretimedoinghousework,preparingfood,andcaringforchildren,
allrelativelyunpopularactivities.Andofcourse,thelargeamountoftimemarriedwomen
spend with their husband is much more pleasant for some than for others. Experienced
well-being is on average unaffected by marriage, not because marriage makes no
difference to happiness but because it changes some aspects of life for the better and
othersfortheworse.
One reason for the low correlations between individuals’ circumstances and their
satisfaction with life is that both experienced happiness and life satisfaction are largely
determinedbythegeneticsoftemperament.Adispositionforwell-beingisasheritableas
heightorintelligence,asdemonstratedbystudiesoftwinsseparatedatbirth.Peoplewho
appearequallyfortunatevarygreatlyinhowhappytheyare.Insomeinstances,asinthe
caseof marriage,the correlations withwell-being are lowbecause ofbalancing effects.
The same situation may be good for some people and bad for others, and new
circumstances have both benefits and costs. In other cases, such as high income, the
effectsonlifesatisfactionaregenerallypositive,butthepictureiscomplicatedbythefact
thatsomepeoplecaremuchmoreaboutmoneythanothersdo.
A large-scale study of the impact of higher education, which was conducted for
JghtA5aorJghtanotherpurpose,revealedstrikingevidenceofthelifelongeffectsofthe
goals that young people set for themselves. The relevant data were drawn from
questionnairescollectedin1995–1997fromapproximately12,000peoplewhohadstarted
theirhighereducationineliteschoolsin1976.Whentheywere17or18,theparticipants
had filled out a questionnaire in which they rated the goal of “being very well-off
financially” on a 4-point scale ranging from “not important” to “essential.” The
questionnaire they completed twenty years later included measures of their income in
1995,aswellasaglobalmeasureoflifesatisfaction.
Goals make a large difference. Nineteen years after they stated their financial
aspirations,manyofthepeoplewhowantedahighincomehadachievedit.Amongthe
597 physicians and other medical professionals in the sample, for example, each
additionalpointonthemoney-importancescalewasassociatedwithanincrementofover
$14,000ofjobincomein1995dollars!Nonworkingmarriedwomenwerealsolikelyto
havesatisfiedtheirfinancialambitions.Eachpointonthescaletranslatedintomorethan
$12,000ofaddedhouseholdincomeforthesewomen,evidentlythroughtheearningsof
theirspouse.
The importance that people attached to income at age 18 also anticipated their
satisfactionwiththeirincomeasadults.Wecomparedlifesatisfactioninahigh-income
group(morethan$200,000householdincome)toalow-tomoderate-incomegroup(less
than$50,000).Theeffectofincomeonlifesatisfactionwaslargerforthosewhohadlisted
being well-off financially as an essential goal: .57 point on a 5-point scale. The
correspondingdifferenceforthosewhohadindicatedthatmoneywasnotimportantwas
only.12.Thepeoplewhowantedmoneyandgotitweresignificantlymoresatisfiedthan
average;thosewhowantedmoneyanddidn’tgetitweresignificantlymoredissatisfied.
The same principle applies to other goals—one recipe for a dissatisfied adulthood is
settinggoalsthatareespeciallydifficulttoattain.Measuredbylifesatisfaction20years
later, the least promising goal that a young person could have was “becoming
accomplished in a performing art.” Teenagers’ goals influence what happens to them,
wheretheyendup,andhowsatisfiedtheyare.
In part because of these findings I have changed my mind about the definition of
well-being.Thegoalsthatpeoplesetforthemselvesaresoimportanttowhattheydoand
howtheyfeelaboutitthatanexclusivefocusonexperiencedwell-beingisnottenable.We
cannotholdaconceptofwell-beingthatignoreswhatpeoplewant.Ontheotherhand,itis
alsotruethataconceptofwell-beingthatignoreshowpeoplefeelastheyliveandfocuses
onlyonhowtheyfeelwhentheythinkabouttheirlifeisalsountenable.Wemustaccept
thecomplexitiesofahybridview,inwhichthewell-beingofbothselvesisconsidered.
TheFocusingIllusion
Wecaninferfromthespeedwithwhichpeoplerespondtoquestionsabouttheirlife,and
fromtheeffectsofcurrentmoodontheirresponses,thattheydonotengageinacareful
examination when they evaluate their life. They must be using heuristics, which are
examples of both substitution and WYSIATI. Although their view of their life was
influenced by a question about dating or by a coin on the copying machine, the
participantsinthesestudiesdidnotforgetthatthereismoretolifethandatingorfeeling
lucky.Theconceptofhappinessisnotsuddenlychangedbyfindingadime,butSystem1
readily substitutes a small part of it for the whole of it. Any aspect of life to which
attentionisdirectedwillloomJghtA5aoomJghtlargeinaglobalevaluation.Thisisthe
essenceofthefocusingillusion,whichcanbedescribedinasinglesentence:
Nothinginlifeisasimportantasyouthinkitiswhenyouarethinkingaboutit.
TheoriginofthisideawasafamilydebateaboutmovingfromCaliforniatoPrinceton,in
which my wife claimed that people are happier in California than on the East Coast. I
argued that climate is demonstrably not an important determinant of well-being—the
Scandinaviancountriesareprobablythehappiestintheworld.Iobservedthatpermanent
lifecircumstanceshavelittleeffectonwell-beingandtriedinvaintoconvincemywife
that her intuitions about the happiness of Californians were an error of affective
forecasting.
A short time later, with this debate still on my mind, I participated in a workshop
aboutthesocialscienceofglobalwarming.Acolleaguemadeanargumentthatwasbased
on his view of the well-being of the population of planet Earth in the next century. I
arguedthatitwaspreposteroustoforecastwhatitwouldbeliketoliveonawarmerplanet
whenwedidnotevenknowwhatitisliketoliveinCalifornia.Soonafterthatexchange,
mycolleagueDavidSchkadeandIweregrantedresearchfundstostudytwoquestions:
ArepeoplewholiveinCaliforniahappierthanothers?andWhatarethepopularbeliefs
abouttherelativehappinessofCalifornians?
WerecruitedlargesamplesofstudentsatmajorstateuniversitiesinCalifornia,Ohio,
andMichigan.Fromsomeofthemweobtainedadetailedreportoftheirsatisfactionwith
variousaspectsoftheirlives.Fromothersweobtainedapredictionofhowsomeone“with
yourinterestsandvalues”wholivedelsewherewouldcompletethesamequestionnaire.
Asweanalyzedthedata,itbecameobviousthatIhadwonthefamilyargument.As
expected,thestudentsinthetworegionsdifferedgreatlyintheirattitudetotheirclimate:
theCaliforniansenjoyedtheirclimateandtheMidwesternersdespisedtheirs.Butclimate
was not an important determinant of well-being. Indeed, there was no difference
whatsoeverbetweenthelifesatisfactionofstudentsinCaliforniaandintheMidwest.We
alsofoundthatmywifewasnotaloneinherbeliefthatCaliforniansenjoygreaterwell-
beingthanothers.Thestudentsinbothregionsshared the samemistakenview,andwe
wereableto tracetheirerror toan exaggerated beliefin the importanceof climate. We
describedtheerrorasafocusingillusion.
The essence of the focusing illusion is WYSIATI, giving too much weight to the
climate,toolittletoalltheotherdeterminantsofwell-being.Toappreciatehowstrongthis
illusionis,takeafewsecondstoconsiderthequestion:
Howmuchpleasuredoyougetfromyourcar?
Ananswercametoyourmindimmediately;youknowhowmuchyoulikeandenjoyyour
car.Nowexamineadifferentquestion:“Whendoyougetpleasurefromyourcar?”The
answertothisquestionmaysurpriseyou,butitisstraightforward:you getpleasure(or
displeasure) from your car when you think about your car, which is probably not very
often.Undernormalcircumstances,youdonotspendmuchtimethinkingaboutyourcar
when you are driving it. You think of other things as you drive, and your mood is
determinedbywhateveryouthinkabout.Hereagain,whenyoutriedtoratehowmuch
youenjoyedyourcar,youactuallyansweredJghtA5aedJghtamuchnarrowerquestion:
“Howmuchpleasuredoyougetfromyourcarwhenyouthinkaboutit?”Thesubstitution
caused you to ignore the fact that you rarely think about your car, a form of duration
neglect.Theupshotisafocusingillusion.Ifyoulikeyourcar,youarelikelytoexaggerate
thepleasureyouderivefromit,whichwillmisleadyouwhenyouthinkofthevirtuesof
yourcurrentvehicleaswellaswhenyoucontemplatebuyinganewone.
AsimilarbiasdistortsjudgmentsofthehappinessofCalifornians.Whenaskedabout
thehappinessofCalifornians,youprobablyconjureanimageofsomeoneattendingtoa
distinctiveaspectoftheCaliforniaexperience,suchashikinginthesummeroradmiring
themildwinterweather.ThefocusingillusionarisesbecauseCaliforniansactuallyspend
little time attending to these aspects of their life. Moreover, long-term Californians are
unlikelytoberemindedoftheclimatewhenaskedforaglobalevaluationoftheirlife.If
youhavebeenthereallyourlifeanddonottravelmuch,livinginCaliforniaislikehaving
tentoes:nice,butnotsomethingonethinksmuchabout.Thoughtsofanyaspectoflifeare
morelikelytobesalientifacontrastingalternativeishighlyavailable.
People who recently moved to California will respond differently. Consider an
enterprisingsoulwhomovedfromOhiotoseekhappinessinabetterclimate.Forafew
yearsfollowingthemove,aquestionabouthissatisfactionwithlifewillprobablyremind
himofthemoveandalsoevokethoughtsofthecontrastingclimatesinthetwostates.The
comparisonwillsurelyfavorCalifornia,andtheattentiontothataspectoflifemaydistort
its true weight in experience. However, the focusing illusion can also bring comfort.
Whetheror notthe individual isactually happierafterthe move,he willreport himself
happier,becausethoughtsoftheclimatewillmakehimbelievethatheis.Thefocusing
illusioncancausepeopletobewrongabouttheirpresentstateofwell-beingaswellas
aboutthehappinessofothers,andabouttheirownhappinessinthefuture.
Whatproportionofthedaydoparaplegicsspendinabadmood?
Thisquestionalmostcertainlymadeyouthinkofaparaplegicwhoiscurrentlythinking
about some aspect of his condition. Your guess about a paraplegic’s mood is therefore
likelytobeaccurateintheearlydaysafteracripplingaccident;forsometimeafterthe
event,accidentvictimsthinkoflittleelse.Butovertime,withfewexceptions,attentionis
withdrawn from a new situation as it becomes more familiar. The main exceptions are
chronicpain,constantexposuretoloudnoise,andseveredepression.Painandnoiseare
biologically set to be signals that attract attention, and depression involves a self-
reinforcing cycle of miserable thoughts. There is therefore no adaptation to these
conditions.Paraplegia,however,isnotoneoftheexceptions:detailedobservationsshow
thatparaplegicsareinafairlygoodmoodmorethanhalfofthetimeasearlyasonemonth
followingtheiraccident—though their moodiscertainly somberwhenthey thinkabout
theirsituation.Mostofthetime,however,paraplegicswork,read,enjoyjokesandfriends,
andgetangrywhentheyreadaboutpoliticsinthenewspaper.Whentheyareinvolvedin
anyoftheseactivities,theyarenotmuchdifferentfromanyoneelse,andwecanexpect
theexperiencedwell-beingofparaplegicstobenearnormalmuchofthetime.Adaptation
toanewsituation,whethergood or bad,consistsinlargepartof thinking lessandless
about it. In that sense, most long-term circumstances of life, including paraplegia and
marriage,arepart-timestatesthatoneinhabitsonlywhenoneatJghtA5aatJghttendsto
them.
One of the privileges of teaching at Princeton is the opportunity to guide bright
undergraduatesthrougharesearchthesis.Andoneofmyfavoriteexperiencesinthisvein
wasaprojectinwhichBeruriaCohncollectedandanalyzeddatafromasurveyfirmthat
askedrespondentstoestimatetheproportionoftimethatparaplegicsspendinabadmood.
Shesplitherrespondentsintotwogroups:someweretoldthatthecripplingaccidenthad
occurred a month earlier, some a year earlier. In addition, each respondent indicated
whetherheorsheknewaparaplegicpersonally.Thetwogroupsagreedcloselyintheir
judgmentabouttherecentparaplegics:thosewhoknewaparaplegicestimated75%bad
mood; those who had to imagine a paraplegic said 70%. In contrast, the two groups
differedsharplyintheirestimatesofthemoodofparaplegicsayearaftertheaccidents:
thosewhoknewaparaplegicoffered41%astheirestimateofthetimeinthatbadmood.
The estimates of those who were not personally acquainted with a paraplegic averaged
68%. Evidently, those who knew a paraplegic had observed the gradual withdrawal of
attentionfromthecondition,butothersdidnotforecastthatthisadaptationwouldoccur.
Judgments about the mood of lottery winners one month and one year after the event
showedexactlythesamepattern.
Wecanexpectthelifesatisfactionofparaplegicsandthoseafflictedbyotherchronic
andburdensomeconditionstobelowrelativetotheirexperiencedwell-being,becausethe
requesttoevaluatetheirliveswillinevitablyremindthemofthelifeofothersandofthe
lifetheyusedtolead.Consistentwiththisidea,recentstudiesofcolostomypatientshave
produceddramaticinconsistenciesbetweenthepatients’experiencedwell-beingandtheir
evaluations of their lives. Experience sampling shows no difference in experienced
happinessbetweenthesepatientsandahealthypopulation.Yetcolostomypatientswould
be willing to trade away years of their life for a shorter life without the colostomy.
Furthermore, patients whose colostomy has been reversed remember their time in this
conditionasawful,andtheywouldgiveupevenmoreoftheirremaininglifenottohave
toreturntoit.Hereitappearsthattherememberingselfissubjecttoamassivefocusing
illusionaboutthelifethattheexperiencingselfenduresquitecomfortably.
DanielGilbertandTimothyWilsonintroducedthewordmiswantingtodescribebad
choices that arise from errors of affective forecasting. This word deserves to be in
everyday language. The focusing illusion (which Gilbert and Wilson call focalism) is a
rich source of miswanting. In particular, it makes us prone to exaggerate the effect of
significantpurchasesorchangedcircumstancesonourfuturewell-being.
Compare two commitments that will change some aspects of your life: buying a
comfortablenewcarandjoiningagroupthatmeetsweekly,perhapsapokerorbookclub.
Bothexperienceswillbenovelandexcitingatthestart.Thecrucialdifferenceisthatyou
willeventuallypaylittleattentiontothecarasyoudriveit,butyouwillalwaysattendto
thesocialinteractiontowhichyoucommittedyourself.ByWYSIATI,youarelikelyto
exaggerate the long-term benefits of the car, but you are not likely to make the same
mistake for a social gathering or for inherently attention-demanding activities such as
playingtennisorlearningtoplaythecello.Thefocusingillusioncreatesabiasinfavorof
goods and experiences that are initially exciting, even if they will eventually lose their
appeal.Timeisneglected,causingexperiencesthatwillretaintheirattentionvalueinthe
longtermtobeappreciatedlessthantheydeservetobe.
TimeandTimeAgain
Theroleoftimehasbeenarefraininthispartofthebook.Itislogicaltodescribethelife
oftheexperiencingselfasaseriesofmoments,eachwithavalue.Thevalueofanepisode
—Ihavecalleditahedonimetertotal—issimplythesumofthevaluesofitsmoments.
But this is not how the mind represents episodes. The remembering self, as I have
describedit,alsotellsstoriesandmakeschoices,andneitherthestoriesnorthechoices
properlyrepresenttime.Instorytellingmode,anepisodeisrepresentedbyafewcritical
moments,especiallythebeginning,thepeak,andtheend.Durationisneglected.Wesaw
thisfocusonsingularmomentsbothinthecold-handsituationandinVioletta’sstory.
We saw a different form of duration neglect in prospect theory, in which a state is
representedbythetransitiontoit.Winningalotteryyieldsanewstateofwealththatwill
endureforsometime,butdecisionutilitycorrespondstotheanticipatedintensityofthe
reactiontothenewsthatonehaswon.Thewithdrawalofattentionandotheradaptations
tothenewstateareneglected,asonlythatthinsliceoftimeisconsidered.Thesamefocus
onthetransitiontothenewstateandthesameneglectoftimeandadaptationarefoundin
forecastsofthereactiontochronicdiseases,andofcourseinthefocusingillusion.The
mistakethatpeoplemakeinthefocusingillusioninvolvesattentiontoselectedmoments
andneglectofwhathappensatothertimes.Themindisgoodwithstories,butitdoesnot
appeartobewelldesignedfortheprocessingoftime.
Duringthelasttenyearswehavelearnedmanynewfactsabouthappiness.Butwe
havealsolearnedthatthewordhappinessdoesnothaveasimplemeaningandshouldnot
beusedasifitdoes.Sometimesscientificprogressleavesusmorepuzzledthanwewere
before.
SpeakingofThinkingAboutLife
“Shethoughtthatbuyingafancycarwouldmakeherhappier,butitturnedouttobe
anerrorofaffectiveforecasting.”
“Hiscarbrokedownonthewaytoworkthismorningandhe’sinafoulmood.This
isnotagooddaytoaskhimabouthisjobsatisfaction!”
“Shelooksquitecheerfulmostofthetime,butwhensheisaskedshesayssheisvery
unhappy.Thequestionmustmakeherthinkofherrecentdivorce.”
“Buying a larger house may not make us happier in the long term. We could be
sufferingfromafocusingillusion.”
“He has chosen to split his time between two cities. Probably a serious case of
miswanting.”
P
Conclusions
Ibeganthisbookbyintroducingtwofictitiouscharacters,spentsometimediscussingtwo
species,andendedwithtwoselves.ThetwocharactersweretheintuitiveSystem1,which
doesJghtA5`؇J5thefastthinking,andtheeffortfulandslowerSystem2,whichdoesthe
slowthinking,monitorsSystem1,andmaintainscontrolasbestitcanwithinitslimited
resources.ThetwospecieswerethefictitiousEcons,wholiveinthelandoftheory,and
theHumans,whoactintherealworld.Thetwoselvesaretheexperiencingself,which
doestheliving,andtherememberingself,whichkeepsscoreandmakesthechoices.In
this final chapter I consider some applications of the three distinctions, taking them in
reverseorder.
TwoSelves
The possibility of conflicts between the remembering self and the interests of the
experiencingselfturned outtobe aharder problem thanI initially thought.In an early
experiment, the cold-hand study, the combination of duration neglect and the peak-end
rule led to choices that were manifestly absurd. Why would people willingly expose
themselvesto unnecessary pain?Our subjectsleft thechoice to theirremembering self,
preferringtorepeatthetrialthatleftthebettermemory,althoughitinvolvedmorepain.
Choosingby thequalityof thememory maybe justifiedinextreme cases,for example
whenpost-traumaticstressisapossibility,butthecold-handexperiencewasnottraumatic.
Anobjectiveobservermakingthechoiceforsomeoneelsewouldundoubtedlychoosethe
shortexposure,favoringthesufferer’sexperiencingself.Thechoicesthatpeoplemadeon
theirownbehalfarefairlydescribedasmistakes.Durationneglectandthepeak-endrule
intheevaluationofstories,bothattheoperaandinjudgmentsofJen’slife,areequally
indefensible.Itdoesnotmakesensetoevaluateanentirelifebyitslastmoments,orto
givenoweighttodurationindecidingwhichlifeismoredesirable.
TherememberingselfisaconstructionofSystem2.However,thedistinctivefeatures
of the way it evaluates episodes and lives are characteristics of our memory. Duration
neglectandthepeak-endruleoriginateinSystem1anddonotnecessarilycorrespondto
thevaluesofSystem2.Webelievethatdurationisimportant,butourmemorytellsusitis
not.Therulesthatgoverntheevaluationofthepastarepoorguidesfordecisionmaking,
because time does matter. The central fact of our existence is that time is the ultimate
finite resource, but the remembering self ignores that reality. The neglect of duration
combinedwiththepeak-endrule causes abiasthatfavors a shortperiodofintense joy
overalongperiodofmoderatehappiness.Themirrorimageofthesamebiasmakesus
fear a short period of intense but tolerable suffering more than we fear a much longer
periodofmoderatepain.Durationneglectalsomakesuspronetoacceptalongperiodof
mildunpleasantnessbecausetheendwillbebetter,anditfavorsgivingupanopportunity
foralonghappyperiodifitislikelytohaveapoorending.Todrivethesameideatothe
pointofdiscomfort,consider thecommonadmonition,“Don’tdoit,youwillregretit.”
Theadvicesoundswisebecauseanticipatedregretistheverdictoftherememberingself
and we are inclined to accept such judgments as final and conclusive. We should not
forget, however, that the perspective of the remembering self is not always correct. An
objectiveobserverofthehedonimeterprofile,withtheinterestsoftheexperiencingselfin
mind,mightwellofferdifferent advice.Therememberingself’s neglect ofduration,its
exaggerated emphasis on peaks and ends, and its susceptibility to hindsight combine to
yielddistortedreflectionsofouractualexperience.
Incontrast,theduration-weightedconceptionofwell-beingtreatsallmomentsoflife
alike,memorableornot.Somemomentsendupweightedmorethanothers,eitherbecause
theyarememorableSareevaorbecause they areimportant.Thetimethatpeoplespend
dwellingonamemorablemomentshouldbeincludedinitsduration,addingtoitsweight.
Amomentcanalsogainimportancebyalteringtheexperienceofsubsequentmoments.
For example, an hour spent practicing the violin may enhance the experience of many
hoursofplayingorlisteningtomusicyearslater.Similarly,abriefawfuleventthatcauses
PTSDshouldbeweightedbythetotaldurationofthelong-termmiseryitcauses.Inthe
duration-weighted perspective, we can determine only after the fact that a moment is
memorable or meaningful. The statements “I will always remember…” or “this is a
meaningfulmoment”shouldbetakenaspromisesorpredictions,whichcanbefalse—and
oftenare—evenwhenutteredwithcompletesincerity.Itisagoodbetthatmanyofthe
thingswesaywewillalwaysrememberwillbelongforgottentenyearslater.
Thelogicofdurationweightingiscompelling,butitcannotbeconsideredacomplete
theory of well-being because individuals identify with their remembering self and care
about their story. A theory of well-being that ignores what people want cannot be
sustained.Ontheotherhand,atheorythatignoreswhatactuallyhappensinpeople’slives
and focuses exclusively on what they think about their life is not tenable either. The
remembering self and the experiencing self must both be considered, because their
interestsdo not alwayscoincide. Philosopherscould strugglewiththese questionsfor a
longtime.
The issue of which of the two selves matters more is not a question only for
philosophers; it has implications for policies in several domains, notably medicine and
welfare.Considertheinvestmentthatshouldbemadeinthetreatmentofvariousmedical
conditions, including blindness, deafness, or kidney failure. Should the investments be
determinedbyhowmuchpeoplefeartheseconditions?Shouldinvestmentsbeguidedby
thesufferingthatpatientsactuallyexperience?Orshouldtheyfollowtheintensityofthe
patients’desiretoberelievedfromtheirconditionandbythesacrificesthattheywouldbe
willing to make to achieve that relief? The ranking of blindness and deafness, or of
colostomy and dialysis, might well be different depending on which measure of the
severityofsufferingisused.Noeasysolutionisinsight,buttheissueistooimportantto
beignored.
The possibility of using measures of well-being as indicators to guide government
policieshasattractedconsiderablerecentinterest,bothamongacademicsandin several
governmentsinEurope.Itisnowconceivable,asitwasnotevenafewyearsago,thatan
indexoftheamountofsufferinginsocietywillsomedaybeincludedinnationalstatistics,
alongwithmeasuresofunemployment,physicaldisability,andincome.Thisprojecthas
comealongway.
EconsandHumans
Ineverydayspeech,wecallpeoplereasonableifitispossibletoreasonwiththem,iftheir
beliefs are generally in tune with reality, and if their preferences are in line with their
interests and their values. The word rational conveys an image of greater deliberation,
morecalculation,andlesswarmth,butincommonlanguagearationalpersoniscertainly
reasonable.Foreconomistsanddecisiontheorists,theadjectivehasanaltogetherdifferent
meaning.Theonlytestofrationalityisnotwhetheraperson’sbeliefsandpreferencesare
reasonable, but whether they are internally consistent. A rational person can believe in
ghostssolongasallherotherbeliefsareconsistentwiththeexistenceofghosts.Arational
personcanpreferbeinghatedoverbeingloved,solongashiSsoasallspreferencesare
consistent.Rationalityislogicalcoherence—reasonableornot.Econsarerationalbythis
definition,butthereis overwhelming evidencethatHumanscannot be. AnEconwould
notbesusceptibletopriming,WYSIATI,narrowframing,theinsideview,orpreference
reversals,whichHumanscannotconsistentlyavoid.
The definition of rationality as coherence is impossibly restrictive; it demands
adherencetorulesoflogicthatafinitemindisnotabletoimplement.Reasonablepeople
cannotberationalbythatdefinition,buttheyshouldnotbebrandedasirrationalforthat
reason. Irrational is a strong word, which connotes impulsivity, emotionality, and a
stubbornresistancetoreasonableargument.IoftencringewhenmyworkwithAmosis
creditedwithdemonstratingthathumanchoicesareirrational,wheninfactourresearch
onlyshowedthatHumansarenotwelldescribedbytherational-agentmodel.
Although Humans are not irrational, they often need help to make more accurate
judgments and better decisions, and in some cases policies and institutions can provide
thathelp.Theseclaimsmayseeminnocuous,buttheyareinfactquitecontroversial.As
interpretedby the importantChicago schoolof economics,faithin humanrationality is
closely linked to an ideology in which it is unnecessary and even immoral to protect
people against their choices. Rational people should be free, and they should be
responsible for taking care of themselves. Milton Friedman, the leading figure in that
school,expressedthisviewinthetitleofoneofhispopularbooks:FreetoChoose.
The assumption that agents are rational provides the intellectual foundation for the
libertarianapproachtopublicpolicy:donotinterferewiththeindividual’srighttochoose,
unlessthechoicesharmothers.Libertarianpoliciesarefurtherbolsteredbyadmirationfor
theefficiencyofmarketsinallocatinggoodstothepeoplewhoarewillingtopaythemost
for them. A famous example of the Chicago approach is titled A Theory of Rational
Addiction; it explains how a rational agent with a strong preference for intense and
immediate gratification may make the rational decision to accept future addiction as a
consequence.IonceheardGaryBecker,oneoftheauthorsofthatarticle,whoisalsoa
NobellaureateoftheChicagoschool,argueinalightervein,butnotentirelyasajoke,
that we should consider the possibility of explaining the so-called obesity epidemic by
people’s belief that a cure for diabetes will soon become available. He was making a
valuable point: when we observe people acting in ways that seem odd, we should first
examinethe possibilitythattheyhaveagoodreasontodowhatthey do.Psychological
interpretations should only be invoked when the reasons become implausible—which
Becker’sexplanationofobesityprobablyis.
InanationofEcons,governmentshouldkeepoutoftheway,allowingtheEconsto
actastheychoose,solongastheydonotharmothers.Ifamotorcycleriderchoosesto
ridewithoutahelmet,alibertarianwillsupporthisrighttodoso.Citizensknowwhatthey
are doing, even when they choose not to save for their old age, or when they expose
themselves to addictive substances. There is sometimes a hard edge to this position:
elderly people who did not save enough for retirement get little more sympathy than
someonewhocomplainsaboutthebillafterconsumingalargemealatarestaurant.Much
is therefore at stake in the debate between the Chicago school and the behavioral
economists, who reject the extreme form of the rational-agent model. Freedom is not a
contested value; all the participants in the debate are in favor of it. But life is more
complexforbehavioraleconomiststhanfortruSth17;ebelieversinhumanrationality.No
behavioraleconomistfavorsastatethatwillforceitscitizenstoeatabalanceddietandto
watch only television programs that are good for the soul. For behavioral economists,
however,freedomhasacost,whichisbornebyindividualswhomakebadchoices,andby
a society that feels obligated to help them. The decision of whether or not to protect
individualsagainsttheirmistakesthereforepresentsadilemmaforbehavioraleconomists.
TheeconomistsoftheChicagoschooldonotfacethatproblem,becauserationalagentsdo
notmakemistakes.Foradherentsofthisschool,freedomisfreeofcharge.
In2008theeconomistRichardThalerandthejuristCassSunsteinteameduptowrite
a book, Nudge, which quickly became an international bestseller and the bible of
behavioral economics. Their book introduced several new words into the language,
includingEconsandHumans.Italsopresentedasetofsolutionstothedilemmaofhowto
help people make good decisions without curtailing their freedom.Thaler and Sunstein
advocateapositionoflibertarianpaternalism,inwhichthestateandotherinstitutionsare
allowedtonudgepeopletomakedecisionsthatservetheirownlong-terminterests.The
designationofjoiningapensionplanasthedefaultoptionisanexampleofanudge.Itis
difficulttoarguethatanyone’sfreedomisdiminishedbybeingautomaticallyenrolledin
theplan,whentheymerelyhavetocheckaboxtooptout.Aswesawearlier,theframing
oftheindividual’sdecision—ThalerandSunsteincallitchoicearchitecture—hasahuge
effectontheoutcome.Thenudgeisbasedonsoundpsychology,whichIdescribedearlier.
Thedefaultoptionisnaturallyperceivedasthenormalchoice.Deviatingfromthenormal
choiceisanactofcommission,whichrequiresmoreeffortfuldeliberation,takesonmore
responsibility,andismorelikelytoevokeregretthandoingnothing.Thesearepowerful
forcesthatmayguidethedecisionofsomeonewhoisotherwiseunsureofwhattodo.
Humans,morethanEcons,alsoneedprotectionfromotherswhodeliberatelyexploit
their weaknesses—and especially the quirks of System 1 and the laziness of System 2.
Rational agents are assumed to make important decisions carefully, and to use all the
informationthatisprovidedtothem.AnEconwillreadandunderstandthefineprintofa
contractbeforesigningit,butHumansusuallydonot.Anunscrupulousfirmthatdesigns
contractsthatcustomerswillroutinelysignwithoutreadinghasconsiderablelegalleeway
in hiding important information in plain sight. A pernicious implication of the rational-
agent model in its extreme form is that customers are assumed to need no protection
beyondensuringthattherelevantinformationisdisclosed.Thesizeoftheprintandthe
complexityofthelanguageinthedisclosurearenotconsideredrelevant—anEconknows
howtodealwithsmallprintwhenitmatters.Incontrast,therecommendationsofNudge
requirefirmstooffercontractsthataresufficientlysimpletobereadandunderstoodby
Humancustomers.Itisagoodsignthatsomeoftheserecommendationshaveencountered
significantoppositionfromfirmswhoseprofitsmightsufferiftheircustomerswerebetter
informed.Aworldinwhichfirmscompetebyofferingbetterproductsispreferabletoone
inwhichthewinneristhefirmthatisbestatobfuscation.
Aremarkablefeatureoflibertarianpaternalismisitsappealacrossabroadpolitical
spectrum.Theflagship exampleofbehavioral policy,called Save MoreTomorrow,was
sponsoredinCongressbyanunusualcoalitionthatincludedextremeconservativesaswell
asliberals.SaveMoreTomorrowisafinancialplanthatfirmscanoffertheiremployees.
ThosewhosignonallowtheemployertoincreaSyersliberalsetheircontributiontotheir
savingplanbyafixedproportionwhenevertheyreceivearaise.Theincreasedsavingrate
isimplementedautomaticallyuntiltheemployeegivesnoticethatshewantstooptoutof
it.Thisbrilliantinnovation,proposedbyRichardThalerandShlomoBenartziin2003,has
nowimprovedthesavingsrateandbrightenedthefutureprospectsofmillionsofworkers.
Itissoundlybasedinthepsychologicalprinciplesthatreadersofthisbookwillrecognize.
Itavoidstheresistancetoanimmediatelossbyrequiringnoimmediatechange;bytying
increasedsavingtopayraises,itturnslossesintoforegonegains,whicharemucheasierto
bear;andthefeatureofautomaticityalignsthelazinessofSystem2withthelong-term
interestsoftheworkers.Allthis,ofcourse,withoutcompellinganyonetodoanythinghe
doesnotwishtodoandwithoutanymisdirectionorartifice.
The appeal of libertarian paternalism has been recognized in many countries,
includingtheUKandSouthKorea,andbypoliticiansofmanystripes,includingTories
andtheDemocraticadministrationofPresidentObama.Indeed,Britain’sgovernmenthas
createdanewsmallunitwhosemissionistoapplytheprinciplesofbehavioralscienceto
helpthegovernmentbetteraccomplishitsgoals.Theofficialnameforthisgroupisthe
BehaviouralInsightTeam,butitisknownbothinandoutofgovernmentsimplyasthe
NudgeUnit.Thalerisanadvisertothisteam.
In a storybook sequel to the writing of Nudge, Sunstein was invited by President
Obamatoserveasadministratorof the OfficeofInformationandRegulatoryAffairs,a
position that gave him considerable opportunity to encourage the application of the
lessonsofpsychologyandbehavioraleconomicsingovernmentagencies.Themissionis
describedinthe2010ReportoftheOfficeofManagementandBudget.Readersofthis
book will appreciate the logic behind specific recommendations, including encouraging
“clear,simple,salient,andmeaningfuldisclosures.”Theywillalsorecognizebackground
statementssuchas“presentationgreatlymatters;if, for example,a potentialoutcome is
framedasaloss,itmayhavemoreimpactthanifitispresentedasagain.”
The example of a regulation about the framing of disclosures concerning fuel
consumptionwasmentionedearlier.Additionalapplicationsthathavebeenimplemented
includeautomaticenrollmentinhealthinsurance,anewversionofthedietaryguidelines
thatreplacestheincomprehensibleFoodPyramidwiththepowerfulimageofaFoodPlate
loadedwithabalanceddiet,andaruleformulatedbytheUSDAthatpermitstheinclusion
of messages such as “90% fat-free” on the label of meat products, provided that the
statement“10%fat”isalsodisplayed“contiguousto,inletteringofthesamecolor,size,
and type as, and on the same color background as, the statement of lean percentage.”
Humans, unlike Econs, need help to make good decisions, and there are informed and
unintrusivewaystoprovidethathelp.
TwoSystems
Thisbookhasdescribedtheworkingsofthemindasanuneasyinteractionbetweentwo
fictitiouscharacters:theautomaticSystem1andtheeffortfulSystem2.Youarenowquite
familiarwiththepersonalitiesofthetwosystemsandabletoanticipatehowtheymight
respondindifferentsituations.Andofcourseyoualsorememberthatthetwosystemsdo
not really exist in the brain or anywhere else. “System 1 does X” is a shortcut for “X
occurs automatically.” And “System 2 is mobilized to do Y” is a shortcut for “arousal
increases, pupils dilate, attention is fo Stenations,cused, and activity Y is performed.” I
hopeyoufindthelanguageofsystemsashelpfulasIdo,andthatyouhaveacquiredan
intuitivesenseofhowtheyworkwithoutgettingconfusedbythequestionofwhetherthey
exist.Havingdeliveredthisnecessarywarning,Iwillcontinuetousethelanguagetothe
end.
TheattentiveSystem2iswhowethinkweare.System2articulatesjudgmentsand
makeschoices,butitoftenendorsesorrationalizesideasandfeelingsthatweregenerated
bySystem1.Youmaynotknowthatyouareoptimisticaboutaprojectbecausesomething
aboutitsleaderremindsyouofyourbelovedsister,orthatyoudislikeapersonwholooks
vaguely like your dentist. If asked for an explanation, however, you will search your
memoryforpresentablereasonsandwillcertainlyfindsome.Moreover,youwillbelieve
the story you make up. But System 2 is not merely an apologist for System 1; it also
prevents many foolish thoughts and inappropriate impulses from overt expression. The
investmentofattentionimprovesperformanceinnumerousactivities—thinkoftherisks
ofdrivingthroughanarrowspacewhileyourmindiswandering—andisessentialtosome
tasks,includingcomparison,choice,andorderedreasoning.However,System2isnota
paragon of rationality. Its abilities are limited and so is the knowledge to which it has
access.Wedonotalwaysthinkstraightwhenwereason,andtheerrorsarenotalwaysdue
tointrusiveandincorrectintuitions.Oftenwemakemistakesbecausewe(ourSystem2)
donotknowanybetter.
IhavespentmoretimedescribingSystem1,andhavedevotedmanypagestoerrors
of intuitive judgment and choice that I attribute to it. However, the relative number of
pages is a poor indicator of the balance between the marvels and the flaws of intuitive
thinking.System1isindeedtheoriginofmuchthatwedowrong,butitisalsotheorigin
ofmostofwhatwedoright—whichismostofwhatwedo.Ourthoughtsandactionsare
routinelyguidedbySystem1andgenerallyareonthemark.Oneofthemarvelsistherich
anddetailedmodelofourworldthatismaintainedinassociativememory:itdistinguishes
surprisingfromnormaleventsinafractionofasecond,immediatelygeneratesanideaof
what was expected instead of a surprise, and automatically searches for some causal
interpretationofsurprisesandofeventsastheytakeplace.
Memory also holds the vast repertory of skills we have acquired in a lifetime of
practice,whichautomaticallyproduceadequatesolutionstochallengesastheyarise,from
walkingaroundalargestoneonthepathtoavertingtheincipientoutburstofacustomer.
The acquisition of skills requires a regular environment, an adequate opportunity to
practice, and rapid and unequivocal feedback about the correctness of thoughts and
actions. When these conditions are fulfilled, skill eventually develops, and the intuitive
judgmentsandchoicesthatquicklycometomindwillmostlybeaccurate.Allthisisthe
work of System 1, which means it occurs automatically and fast. A marker of skilled
performanceistheabilitytodealwithvastamountsofinformationswiftlyandefficiently.
When a challenge is encountered to which a skilled response is available, that
responseisevoked.Whathappensintheabsenceofskill?Sometimes,asintheproblem
17×24=?,whichcallsforaspecificanswer,itisimmediatelyapparentthatSystem2
must be called in. But it is rare for System 1 to be dumbfounded. System 1 is not
constrained by capacity limits and is profligate in its computations. When engaged in
searchingforananswertoonequestion,itsimultaneouslygeneratestheanswerstorelated
questions,anditmaysubstitutearesponsethatmoreeasilycomestomindfortheonethat
was requested. In this conception of heu Septtedristics, the heuristic answer is not
necessarilysimplerormorefrugalthantheoriginalquestion—itisonlymoreaccessible,
computed more quickly and easily. The heuristic answers are not random, and they are
oftenapproximatelycorrect.Andsometimestheyarequitewrong.
System1registersthecognitiveeasewithwhichitprocessesinformation,butitdoes
notgenerateawarningsignalwhenitbecomesunreliable.Intuitiveanswerscometomind
quicklyandconfidently,whethertheyoriginatefromskillsorfromheuristics.Thereisno
simplewayforSystem2todistinguishbetweenaskilledandaheuristicresponse.Itsonly
recourse is to slow down and attempt to construct an answer on its own, which it is
reluctanttodobecauseitisindolent.ManysuggestionsofSystem1arecasuallyendorsed
withminimalchecking,asinthebat-and-ballproblem.ThisishowSystem1acquiresits
bad reputation as the source of errors and biases. Its operative features, which include
WYSIATI, intensity matching, and associative coherence, among others, give rise to
predictablebiasesandtocognitiveillusionssuchasanchoring,nonregressivepredictions,
overconfidence,andnumerousothers.
Whatcanbedoneaboutbiases?Howcanweimprovejudgmentsanddecisions,both
ourownandthoseoftheinstitutionsthatweserveandthatserveus?Theshortansweris
thatlittle canbe achieved without a considerableinvestment ofeffort.As Iknow from
experience, System 1 is not readily educable. Except for some effects that I attribute
mostly to age, my intuitive thinking is just as prone to overconfidence, extreme
predictions,andtheplanningfallacyasitwasbeforeImadeastudyoftheseissues.Ihave
improved only in my ability to recognize situations in which errors are likely: “This
numberwillbeananchor…,”“Thedecisioncouldchangeiftheproblemisreframed…”
AndIhavemademuchmoreprogressinrecognizingtheerrorsofothersthanmyown.
ThewaytoblockerrorsthatoriginateinSystem1issimpleinprinciple:recognize
thesignsthatyouareinacognitiveminefield,slowdown,andaskforreinforcementfrom
System2.ThisishowyouwillproceedwhenyounextencountertheMüller-Lyerillusion.
When you see lines with fins pointing in different directions, you will recognize the
situationasoneinwhichyoushouldnottrustyourimpressionsoflength.Unfortunately,
thissensibleprocedureisleastlikelytobeappliedwhenitisneededmost.Wewouldall
like to have a warning bell that rings loudly whenever we are about to make a serious
error,butnosuchbellisavailable,andcognitiveillusionsaregenerallymoredifficultto
recognizethanperceptualillusions.Thevoiceofreasonmaybemuchfainterthantheloud
and clear voice of an erroneous intuition, and questioning your intuitions is unpleasant
whenyoufacethestressofabigdecision.Moredoubtisthelastthingyouwantwhenyou
are in trouble. The upshot is that it is much easier to identify a minefield when you
observe others wandering into it than when you are about to do so. Observers are less
cognitively busy and more open to information than actors. That was my reason for
writingabookthatisorientedtocriticsandgossipersratherthantodecisionmakers.
Organizationsarebetterthanindividualswhenitcomestoavoidingerrors,because
they naturally think more slowly and have the power to impose orderly procedures.
Organizations can institute and enforce the application of useful checklists, as well as
moreelaborateexercises,suchasreference-classforecastingandthepremortem.Atleast
inpartbyprovidingadistinctivevocabulary,organizationscanalsoencourageaculturein
which people watch out for one another as they approach minefields. Whatever else it
produces, a St pof othersn organization is a factory that manufactures judgments and
decisions.Everyfactorymusthavewaystoensurethequalityofitsproductsintheinitial
design,infabrication,andinfinalinspections.Thecorrespondingstagesintheproduction
ofdecisionsaretheframingoftheproblemthatistobesolved,thecollectionofrelevant
informationleadingtoadecision,andreflectionandreview.Anorganizationthatseeksto
improveitsdecisionproductshouldroutinelylookforefficiencyimprovementsateachof
thesestages.Theoperativeconceptisroutine.Constantqualitycontrolisanalternativeto
thewholesalereviewsofprocessesthatorganizationscommonlyundertakeinthewakeof
disasters. There is much to be done to improve decision making. One example out of
manyistheremarkableabsenceofsystematictrainingfortheessentialskillofconducting
efficientmeetings.
Ultimately,aricherlanguageisessentialtotheskillofconstructivecriticism.Much
likemedicine,theidentificationofjudgmenterrorsisadiagnostictask,whichrequiresa
precisevocabulary.Thenameofadiseaseisahooktowhichallthatisknownaboutthe
diseaseisattached,includingvulnerabilities,environmentalfactors,symptoms,prognosis,
andcare.Similarly,labels such as“anchoringeffects,”“narrowframing,”or“excessive
coherence” bring together in memory everything we know about a bias, its causes, its
effects,andwhatcanbedoneaboutit.
Thereisadirectlinkfrommoreprecisegossipatthewatercoolertobetterdecisions.
Decisionmakersaresometimesbetterabletoimaginethevoicesofpresentgossipersand
futurecriticsthantohearthehesitantvoiceoftheirowndoubts.Theywillmakebetter
choices when they trust their critics to be sophisticated and fair, and when they expect
theirdecisiontobejudgedbyhowitwasmade,notonlybyhowitturnedout.
P
AppendixA:JudgmentUnderUncertainty:Heuristicsand
Biases*
AmosTverskyandDanielKahneman
Manydecisionsarebasedonbeliefsconcerningthelikelihoodofuncertaineventssuchas
theoutcomeofanelection,theguiltofadefendant,orthefuturevalueofthedollar.These
beliefsareusuallyexpressedinstatementssuchas“Ithinkthat…,”“chancesare…,”“itis
unlikely that…,” and so forth. Occasionally, beliefs concerning uncertain events are
expressed in numerical form as odds or subjective probabilities. What determines such
beliefs? How do people assess the probability of an uncertain event or the value of an
uncertainquantity?Thisarticleshows that peoplerelyonalimitednumberofheuristic
principleswhichreducethecomplextasksofassessingprobabilitiesandpredictingvalues
to simpler judgmental operations. In general, these heuristics are quite useful, but
sometimestheyleadtosevereandsystematicerrors.
The subjective assessment of probability resembles the subjective assessment of
physical quantities such as distance or size. These judgments are all based on data of
limited validity, which are processed according to heuristic rules. For example, the
apparentdistanceofanobjectisdeterminedin partbyitsclarity.Themoresharplythe
objectisseen,thecloseritappearstobe.Thisrulehassomevalidity,becauseinanygiven
scene the more distant objects are seen less sharply than Vt pofreak/>stimated when
visibilityisgoodbecausetheobjectsareseensharply.Thus,therelianceonclarityasan
indicationofdistanceleadstocommonbiases.Suchbiasesarealsofoundintheintuitive
judgmentofprobability.Thisarticledescribesthreeheuristicsthatareemployedtoassess
probabilitiesandtopredictvalues.Biasestowhichtheseheuristicsleadareenumerated,
andtheappliedandtheoreticalimplicationsoftheseobservationsarediscussed.
Representativeness
Manyoftheprobabilisticquestionswithwhichpeopleareconcernedbelongtooneofthe
following types: What is the probability that object A belongs to class B? What is the
probabilitythateventAoriginatesfromprocessB?WhatistheprobabilitythatprocessB
will generate event A? In answering such questions, people typically rely on the
representativenessheuristic,inwhichprobabilitiesareevaluatedbythedegreetowhichA
isrepresentativeofB,thatis,bythedegreetowhichAresemblesB.Forexample,whenA
ishighlyrepresentativeofB,theprobabilitythatAoriginatesfromBisjudgedtobehigh.
On the other hand, if A is not similar to B, the probability that A originates from B is
judgedtobelow.
Foranillustrationofjudgmentbyrepresentativeness,consideranindividualwhohas
been described by a former neighbor as follows: “Steve is very shy and withdrawn,
invariablyhelpful,butwithlittleinterestinpeople,orintheworldofreality.Ameekand
tidysoul,hehasaneedfororderandstructure,andapassionfordetail.”Howdopeople
assess the probability that Steve is engaged in a particular occupation from a list of
possibilities(forexample,farmer,salesman,airlinepilot,librarian,orphysician)?Howdo
people order these occupations from most to least likely? In the representativeness
heuristic,theprobabilitythatSteveisalibrarian,forexample,isassessedbythedegreeto
whichheisrepresentativeof,orsimilarto,thestereotypeofalibrarian.Indeed,research
withproblemsofthistypehasshownthatpeopleordertheoccupationsbyprobabilityand
bysimilarityinexactlythesameway.1Thisapproachtothejudgmentofprobabilityleads
to serious errors, because similarity, or representativeness, is not influenced by several
factorsthatshouldaffectjudgmentsofprobability.
Insensitivitytopriorprobabilityofoutcomes.Oneofthefactorsthathavenoeffecton
representativenessbutshouldhaveamajoreffectonprobabilityisthepriorprobability,or
baseratefrequency,oftheoutcomes.InthecaseofSteve,forexample,thefactthatthere
aremanymorefarmersthanlibrariansinthepopulationshouldenterintoanyreasonable
estimateoftheprobabilitythatSteveisalibrarianratherthanafarmer.Considerationsof
base-ratefrequency,however, donot affectthe similarityof Steveto thestereotypes of
librarians and farmers. If people evaluate probability by representativeness, therefore,
priorprobabilitieswillbeneglected.Thishypothesiswastestedinanexperimentwhere
priorprobabilitiesweremanipulated.2Subjectswereshownbriefpersonalitydescriptions
ofseveralindividuals,allegedlysampledatrandomfromagroupof100professionals—
engineers and lawyers. The subjects were asked to assess, for each description, the
probability that it belonged to an engineer rather than to a lawy [hanerser. In one
experimentalcondition,subjectsweretoldthatthegroupfromwhichthedescriptionshad
beendrawnconsistedof70engineersand30lawyers.Inanothercondition,subjectswere
toldthatthegroupconsistedof30engineersand70lawyers.Theoddsthatanyparticular
description belongs to an engineer rather than to a lawyer should be higher in the first
condition,wherethereisamajorityofengineers,thaninthesecondcondition,wherethere
is a majority of lawyers. Specifically, it can be shown by applying Bayes’ rule that the
ratiooftheseoddsshouldbe(.7/.3)2,or5.44,foreachdescription.Inasharpviolationof
Bayes’rule,thesubjectsinthetwoconditionsproducedessentiallythesameprobability
judgments. Apparently, subjects evaluated the likelihood that a particular description
belongedtoanengineerratherthantoalawyerbythedegreetowhichthisdescription
wasrepresentativeofthetwostereotypes,withlittleornoregardforthepriorprobabilities
ofthecategories.
Thesubjectsusedpriorprobabilitiescorrectlywhentheyhadnootherinformation.In
the absence of a personality sketch, they judged the probability that an unknown
individual is an engineer to be .7 and .3, respectively, in the two base-rate conditions.
However,priorprobabilitieswereeffectivelyignoredwhenadescriptionwasintroduced,
even when this description was totally uninformative. The responses to the following
descriptionillustratethisphenomenon:
Dickisa30-year-oldman.Heismarriedwithnochildren.Amanofhighabilityand
highmotivation,hepromisestobequitesuccessfulinhisfield.Heiswelllikedby
hiscolleagues.
This description was intended to convey no information relevant to the question of
whether Dick is an engineer or a lawyer. Consequently, the probability that Dick is an
engineershouldequaltheproportionofengineersinthegroup,asifnodescriptionhad
beengiven.Thesubjects,however,judgedtheprobabilityofDickbeinganengineertobe
.5 regardless of whether the stated proportion of engineers in the group was .7 or .3.
Evidently,peopleresponddifferentlywhengivennoevidenceandwhengivenworthless
evidence. When no specific evidence is given, prior probabilities are properly utilized;
whenworthlessevidenceisgiven,priorprobabilitiesareignored.3
Insensitivitytosamplesize.Toevaluatetheprobabilityofobtainingaparticularresult
in a sample drawn from a specified population, people typically apply the
representativeness heuristic. That is, they assess the likelihood of a sample result, for
example, that the average height in a random sample often men will be 6 feet, by the
similarityofthisresulttothecorrespondingparameter(thatis,totheaverageheightinthe
populationofmen).Thesimilarityofasamplestatistictoapopulationparameterdoesnot
depend on the size of the sample. Consequently, if probabilities are assessed by
representativeness, then the judged probability of a sample statistic will be essentially
independentofsamplesize.Indeed,whensubjectsassessed the distributionsof average
heightforsamplesofvarioussizes,theyproducedidenticaldistributions.Forexample,the
probabilityofobtaininganaverageheightgreaterthan6feetwasassignedthesamevalue
forsamplesof1,000,100,and10men.4Moreover,subjectsfailedtoappreciatetheroleof
samplesizeevenwhenitwasemphasizedintheformulationoftheproblem.Considerthe
followingquestion:
Acertaintowniss[ainquotewiervedbytwohospitals.Inthelargerhospitalabout45
babiesareborneachday,andinthesmallerhospitalabout15babiesareborneach
day.Asyouknow,about50%ofallbabiesareboys.However,theexactpercentage
variesfromdaytoday.
Sometimesitmaybehigherthan50%,sometimeslower.
Foraperiodof1year,eachhospitalrecordedthedaysonwhichmorethan60%
ofthebabiesbornwereboys.Whichhospitaldoyouthinkrecordedmoresuchdays?
Thelargerhospital(21)
Thesmallerhospital(21)
Aboutthesame(thatis,within5%ofeachother)(53)
The values in parentheses are the number of undergraduate students who chose each
answer.
Mostsubjectsjudgedtheprobabilityofobtainingmorethan60%boystobethesame
inthesmallandinthelargehospital,presumablybecausetheseeventsaredescribedby
the same statistic and are therefore equally representative of the general population. In
contrast,samplingtheory entailsthatthe expected numberof days onwhich morethan
60%of thebabies areboys ismuch greaterin thesmall hospitalthan inthe largeone,
because a large sample is less likely to stray from 50%. This fundamental notion of
statisticsisevidentlynotpartofpeople’srepertoireofintuitions.
A similar insensitivity to sample size has been reported in judgments of posterior
probability,thatis,oftheprobabilitythatasamplehasbeendrawnfromonepopulation
ratherthanfromanother.Considerthefollowingexample:
Imagineanurnfilledwithballs,ofwhich2/3areofonecolorand1/3ofanother.One
individualhasdrawn5ballsfromtheurn,andfoundthat4wereredand1waswhite.
Anotherindividualhasdrawn20ballsandfoundthat12wereredand8werewhite.
Whichofthetwoindividualsshouldfeelmoreconfidentthattheurncontains2/3red
ballsand1/3whiteballs,ratherthantheopposite?Whatoddsshouldeachindividual
give?
Inthisproblem,thecorrectposterioroddsare8to1forthe4:1sampleand16to1forthe
12:8sample,assumingequalpriorprobabilities.However,mostpeoplefeelthatthefirst
sampleprovidesmuchstrongerevidenceforthehypothesisthattheurnispredominantly
red,because the proportion of redballs islargerin thefirst thanin the second sample.
Here again, intuitive judgments are dominated by the sample proportion and are
essentially unaffected by the size of the sample, which plays a crucial role in the
determination of the actual posterior odds.5 In addition, intuitive estimates of posterior
oddsare farless extreme thanthe correctvalues. The underestimationof theimpact of
evidence has been observed repeatedly in problems of this type.6 It has been labeled
“conservatism.”
Misconceptions of chance. People expect that a sequence of events generated by a
randomprocesswillrepresenttheessentialcharacteristicsofthatprocessevenwhenthe
sequenceisshort.Inconsideringtossesofacoinforheadsortails,forexample,people
regardthesequenceH-T-H-T-T-HtobemorelikelythanthesequenceH-H-H-T-[enc.IT-
T,whichdoesnotappearrandom,andalsomorelikelythanthesequenceH-H-H-H-T-H,
whichdoesnotrepresentthefairnessofthecoin.7Thus,peopleexpectthattheessential
characteristicsoftheprocesswillberepresented,notonlygloballyintheentiresequence,
butalsolocallyineachofitsparts.Alocallyrepresentativesequence,however,deviates
systematically from chance expectation: it contains too many alternations and too few
runs. Another consequence of the belief in local representativeness is the well-known
gambler’sfallacy.Afterobservingalongrunofredontheroulettewheel,forexample,
most people erroneously believe that black is now due, presumably because the
occurrenceofblackwillresultinamorerepresentativesequencethantheoccurrenceofan
additional red. Chance is commonly viewed as a self-correcting process in which a
deviation in one direction induces a deviation in the opposite direction to restore the
equilibrium.Infact,deviationsarenot“corrected”asachanceprocessunfolds,theyare
merelydiluted.
Misconceptionsofchancearenotlimitedtonaivesubjects.Astudyofthestatistical
intuitionsofexperiencedresearchpsychologists8revealedalingeringbeliefinwhatmay
becalledthe“lawofsmallnumbers,”accordingtowhichevensmallsamplesarehighly
representative of the populations from which they are drawn. The responses of these
investigatorsreflectedtheexpectationthatavalidhypothesisaboutapopulationwillbe
representedbyastatisticallysignificantresultinasamplewithlittleregardforitssize.As
a consequence, the researchers put too much faith in the results of small samples and
grosslyoverestimatedthereplicabilityofsuchresults.Intheactualconductofresearch,
thisbiasleadstotheselectionofsamplesofinadequatesizeandtooverinterpretationof
findings.
Insensitivity to predictability. People are sometimes called upon to make such
numericalpredictionsasthefuturevalueofastock,thedemandforacommodity,orthe
outcomeofafootballgame.Suchpredictionsareoftenmadebyrepresentativeness.For
example,supposeoneisgivenadescriptionofacompanyandisaskedtopredictitsfuture
profit.Ifthedescriptionofthecompanyisveryfavorable,averyhighprofitwillappear
most representative of that description; if the description is mediocre, a mediocre
performance will appear most representative. The degree to which the description is
favorableisunaffectedbythereliabilityofthatdescriptionorbythedegreetowhichit
permitsaccurateprediction.Hence,ifpeoplepredictsolelyintermsofthefavorableness
ofthedescription,theirpredictionswillbeinsensitivetothereliabilityoftheevidenceand
totheexpectedaccuracyoftheprediction.
This mode of judgment violates the normative statistical theory in which the
extremenessandtherangeofpredictionsarecontrolledbyconsiderationsofpredictability.
Whenpredictabilityisnil,thesamepredictionshouldbemadeinallcases.Forexample,if
the descriptions of companies provide no information relevant to profit, then the same
value(suchasaverageprofit) should bepredictedforallcompanies.If predictability is
perfect, of course, the values predicted will match the actual values and the range of
predictionswillequaltherangeofoutcomes.Ingeneral,thehigherthepredictability,the
widertherangeofpredictedvalues.
Severalstudiesof numerical predictionhavedemonstrated that intuitivepredictions
violate this rule, and that subjects show little or no regard for considerations of
predictability.9 In one o [pand tf these studies, subjects were presented with several
paragraphs, each describing the performance of a student teacher during a particular
practicelesson.Somesubjectswereaskedtoevaluatethequalityofthelessondescribed
in the paragraph in percentile scores, relative to a specified population. Other subjects
were asked to predict, also in percentile scores, the standing of each student teacher 5
years after the practice lesson. The judgments made under the two conditions were
identical.Thatis,thepredictionofaremotecriterion(successofateacherafter5years)
wasidenticaltotheevaluationoftheinformationonwhichthepredictionwasbased(the
qualityofthepracticelesson).Thestudentswhomadethesepredictionswereundoubtedly
awareof thelimitedpredictability ofteaching competence onthe basisof asingle trial
lesson5yearsearlier;nevertheless,theirpredictionswereasextremeastheirevaluations.
The illusion of validity. As we have seen, people often predict by selecting the
outcome (for example, an occupation) that is most representative of the input (for
example, the description of a person). The confidence they have in their prediction
dependsprimarilyonthedegreeofrepresentativeness(thatis,onthequalityofthematch
betweentheselectedoutcomeandtheinput)withlittleornoregardforthefactorsthat
limitpredictiveaccuracy.Thus,peopleexpressgreatconfidenceinthepredictionthata
person is a librarian when given a description of his personality which matches the
stereotype of librarians, even if the description is scanty, unreliable, or outdated. The
unwarrantedconfidencewhichisproducedbyagoodfitbetweenthepredictedoutcome
andtheinputinformationmaybecalledtheillusionofvalidity.Thisillusionpersistseven
whenthejudgeis awareofthefactors that limittheaccuracyof his predictions.Itisa
commonobservationthatpsychologistswhoconductselectioninterviewsoftenexperience
considerableconfidenceintheirpredictions,evenwhentheyknowofthevastliterature
thatshowsselectioninterviewstobehighlyfallible.Thecontinuedrelianceontheclinical
interviewforselection,despiterepeateddemonstrationsofitsinadequacy,amplyatteststo
thestrengthofthiseffect.
The internal consistency of a pattern of inputs is a major determinant of one’s
confidence in predictions based on these inputs. For example, people express more
confidenceinpredictingthefinalgradepointaverageofastudentwhosefirst-yearrecord
consistsentirelyofB’sthaninpredictingthegradepointaverageofastudentwhosefirst-
yearrecordincludesmanyA’sandC’s.Highlyconsistentpatternsaremostoftenobserved
whentheinputvariablesarehighlyredundantorcorrelated.Hence,peopletendtohave
great confidence in predictions based on redundant input variables. However, an
elementaryresultinthestatisticsofcorrelationassertsthat,giveninputvariablesofstated
validity,apredictionbasedonseveralsuchinputscanachievehigheraccuracywhenthey
are independent of each other than when they are redundant or correlated. Thus,
redundancyamonginputsdecreasesaccuracyevenasitincreasesconfidence,andpeople
areoftenconfidentinpredictionsthatarequitelikelytobeoffthemark.10
Misconceptionsofregression.Supposealargegroupofchildrenhasbeenexamined
on two equivalent versions of an aptitude test. If one selects ten children from among
thosewhodidbestononeofthetwoversions,hewillusuallyfindtheirperformanceon
thesecondversiontobesomewhatdisappointing.Conversely,ifoneselectstenchildren
fromamongthosewhodidworstononeversion,theywillbefound,ontheaverage,todo
somewhatbetterontheotherversion.Mo[rvstregenerally,considertwovariablesXand
Y which have the same distribution. If one selects individuals whose average X score
deviatesfromthemeanof Xby k units, thenthe average of their Y scores will usually
deviatefromthemeanof Ybylessthank units. Theseobservations illustrateageneral
phenomenon known as regression toward the mean, which was first documented by
Galtonmorethan100yearsago.
Inthenormalcourseoflife,oneencountersmanyinstancesofregressiontowardthe
mean,inthecomparisonoftheheightoffathersandsons,oftheintelligenceofhusbands
and wives, or of the performance of individuals on consecutive examinations.
Nevertheless,peopledonotdevelopcorrectintuitionsaboutthisphenomenon.First,they
donotexpectregressioninmanycontextswhereitisboundtooccur.Second,whenthey
recognizetheoccurrenceofregression,theyofteninventspuriouscausalexplanationsfor
it.11 We suggest that the phenomenon of regression remains elusive because it is
incompatible with the belief that the predicted outcome should be maximally
representativeoftheinput,and,hence,thatthevalueoftheoutcomevariableshouldbeas
extremeasthevalueoftheinputvariable.
Thefailuretorecognizetheimportofregressioncanhaveperniciousconsequences,
asillustratedbythefollowingobservation.12Inadiscussionofflighttraining,experienced
instructorsnotedthatpraiseforanexceptionallysmoothlandingistypicallyfollowedbya
poorer landing on the next try, while harsh criticism after a rough landing is usually
followed by an improvement on the next try. The instructors concluded that verbal
rewardsaredetrimentaltolearning,whileverbalpunishmentsarebeneficial,contraryto
acceptedpsychologicaldoctrine.Thisconclusionisunwarrantedbecauseofthepresence
ofregressiontowardthemean.Asinothercasesofrepeatedexamination,animprovement
will usually follow a poor performance and a deterioration will usually follow an
outstanding performance, even if the instructor does not respond to the trainee’s
achievementonthefirstattempt.Becausetheinstructorshadpraisedtheirtraineesafter
good landings and admonished them after poor ones, they reached the erroneous and
potentiallyharmfulconclusionthatpunishmentismoreeffectivethanreward.
Thus,thefailuretounderstandtheeffectofregressionleadsonetooverestimatethe
effectiveness of punishment and to underestimate the effectiveness of reward. In social
interaction,aswellasintraining,rewardsaretypicallyadministeredwhenperformanceis
good, and punishments are typically administered when performance is poor. By
regressionalone,therefore,behaviorismostlikelytoimproveafterpunishmentandmost
likely to deteriorate after reward. Consequently, the human condition is such that, by
chancealone,oneismostoftenrewardedforpunishingothersandmostoftenpunishedfor
rewardingthem.Peoplearegenerallynotawareofthiscontingency.Infact,theelusive
role of regression in determining the apparent consequences of reward and punishment
seemstohaveescapedthenoticeofstudentsofthisarea.
Availability
Therearesituationsinwhichpeopleassessthefrequencyofaclassortheprobabilityof
an event by the ease with which instances or occurrences can be brought to mind. For
example,onemayassesstheriskofheartattackamongmiddle-agedpeoplebyrecalling
suchoccurrencesa[occpunishmentmongone’sacquaintances.Similarly,onemayevaluate
theprobabilitythatagivenbusinessventurewillfailbyimaginingvariousdifficultiesit
could encounter. This judgmental heuristic is called availability. Availability is a useful
clueforassessingfrequencyorprobability,becauseinstancesoflargeclassesareusually
recalledbetterandfasterthaninstancesoflessfrequentclasses.However,availabilityis
affected by factors other than frequency and probability. Consequently, the reliance on
availabilityleadstopredictablebiases,someofwhichareillustratedbelow.
Biasesduetotheretrievabilityofinstances.Whenthesizeofaclassisjudgedbythe
availabilityofitsinstances,aclasswhoseinstancesareeasilyretrievedwillappearmore
numerous than a class of equal frequency whose instances are less retrievable. In an
elementarydemonstrationofthiseffect,subjectsheardalistofwell-knownpersonalities
ofbothsexesandweresubsequentlyaskedtojudgewhetherthelistcontainedmorenames
ofmenthanofwomen.Differentlistswerepresentedtodifferentgroupsofsubjects.In
someoftheliststhemenwererelativelymorefamousthanthewomen,andinothersthe
women were relatively more famous than the men. In each of the lists, the subjects
erroneously judged that the class (sex) that had the more famous personalities was the
morenumerous.13
In addition to familiarity, there are other factors, such as salience, which affect the
retrievability of instances. For example, the impact of seeing a house burning on the
subjective probability of such accidents is probably greater than the impact of reading
aboutafireinthelocalpaper.Furthermore,recentoccurrencesarelikelytoberelatively
more available than earlier occurrences. It is a common experience that the subjective
probability of traffic accidents rises temporarily when one sees a car overturned by the
sideoftheroad.
Biasesduetotheeffectivenessofasearchset.Supposeonesamplesaword(ofthree
lettersormore)atrandomfromanEnglishtext.Isitmorelikelythatthewordstartswithr
orthatris the thirdletter?Peopleapproachthisproblemby recalling words thatbegin
with r (road) and words that have r in the third position (car) and assess the relative
frequency by the ease with which words of the two types come to mind. Because it is
mucheasiertosearchforwordsbytheirfirstletterthanbytheirthirdletter,mostpeople
judgewordsthatbeginwithagivenconsonanttobemorenumerousthanwordsinwhich
thesameconsonantappearsinthethirdposition.Theydosoevenforconsonants,suchas
rork,thataremorefrequentinthethirdpositionthaninthefirst.14
Differenttaskselicitdifferentsearchsets.Forexample,supposeyouareaskedtorate
thefrequencywithwhichabstractwords(thought,love)andconcretewords(door,water)
appearinwrittenEnglish.Anaturalwaytoanswerthisquestionistosearchforcontexts
inwhichthewordcouldappear.Itseemseasiertothinkofcontextsinwhichanabstract
conceptismentioned(loveinlovestories)thantothinkofcontextsinwhichaconcrete
word(suchasdoor)ismentioned.Ifthefrequencyofwordsisjudgedbytheavailability
of the contexts in which they appear, abstract words will be judged as relatively more
numerous than concrete words. This bias has been observed in a recent study15 which
showedthatthejudgedfrequencyofoccurrenceofabstractwordswasmuchhigherthan
thatofconcretewords,equatedinobjectivefrequency.Abstractwordswerealsojudgedto
appearinamuchgreatervarietyofcontextsthanconcretewords.
Biasesofimaginability.Sometimesonehastoassessthefrequencyofaclasswhose
instancesarenotstoredinmemorybutcanbegeneratedaccordingtoagivenrule.Insuch
situations,onetypicallygeneratesseveralinstancesandevaluatesfrequencyorprobability
bytheease with whichtherelevant instances canbeconstructed. However, theeaseof
constructing instances does not always reflect their actual frequency, and this mode of
evaluation is prone to biases. To illustrate, consider a group of 10 people who form
committeesofkmembers,2=k=8.Howmanydifferentcommitteesofkmemberscanbe
formed? The correct answer to this problem is given by the binomial coefficient (10/k)
which reaches a maximum of 252 for k= 5. Clearly, the number of committees of k
membersequalsthenumberofcommitteesof(10–k)members,becauseanycommittee
ofkmembersdefinesauniquegroupof(10–k)nonmembers.
One way to answer this question without computation is to mentally construct
committeesofkmembersandtoevaluatetheirnumberbytheeasewithwhichtheycome
tomind.Committeesoffewmembers,say2,aremoreavailablethancommitteesofmany
members,say8.Thesimplestschemefortheconstructionofcommitteesisapartitionof
the group into disjoint sets. One readily sees that it is easy to construct five disjoint
committeesof2members,whileitisimpossibletogenerateeventwodisjointcommittees
of8members.Consequently,iffrequencyisassessedbyimaginability,orbyavailability
forconstruction,thesmallcommitteeswillappearmorenumerousthanlargercommittees,
incontrasttothecorrectbell-shapedfunction.Indeed,whennaivesubjectswereaskedto
estimate the number of distinct committees of various sizes, their estimates were a
decreasingmonotonicfunctionofcommitteesize.16Forexample,themedianestimateof
thenumberofcommitteesof2memberswas70,whiletheestimateforcommitteesof8
memberswas20(thecorrectansweris45inbothcases).
Imaginability plays an important role in the evaluation of probabilities in real-life
situations.Theriskinvolvedinanadventurousexpedition,forexample,isevaluatedby
imaginingcontingencieswithwhichtheexpeditionisnotequippedtocope.Ifmanysuch
difficulties are vividly portrayed, the expedition can be made to appear exceedingly
dangerous, although the ease with which disasters are imagined need not reflect their
actual likelihood. Conversely, the risk involved in an undertaking may be grossly
underestimatedifsomepossibledangersareeitherdifficulttoconceiveof,orsimplydo
notcometomind.
Illusorycorrelation.ChapmanandChapman17havedescribedaninterestingbiasin
the judgment of the frequency with which two events co-occur. They presented naive
judges with information concerning several hypothetical mental patients. The data for
each patient consisted of a clinical diagnosis and a drawing of a person made by the
patient. Later the judges estimated the frequency with which each diagnosis (such as
paranoia or suspiciousness) had been accompanied by various features of the drawing
(suchaspeculiareyes).Thesubjectsmarkedlyoverestimatedthefrequencyof[frpicico-
occurrenceofnaturalassociates,suchassuspiciousnessandpeculiareyes.Thiseffectwas
labeled illusory correlation. In their erroneous judgments of the data to which they had
beenexposed,naivesubjects“rediscovered”muchofthecommon,butunfounded,clinical
loreconcerningtheinterpretationofthedraw-a-persontest.Theillusorycorrelationeffect
was extremely resistant to contradictory data. It persisted even when the correlation
betweensymptomanddiagnosiswasactuallynegative,anditpreventedthejudgesfrom
detectingrelationshipsthatwereinfactpresent.
Availability provides a natural account for the illusory-correlation effect. The
judgment of how frequently two events co-occur could be based on the strength of the
associativebondbetweenthem.Whentheassociationisstrong,oneislikelytoconclude
thattheeventshavebeenfrequentlypaired.Consequently,strongassociateswillbejudged
to have occurred together frequently. According to this view, the illusory correlation
betweensuspiciousnessandpeculiardrawingoftheeyes,forexample,isduetothefact
thatsuspiciousnessismorereadilyassociatedwiththeeyesthanwithanyotherpartofthe
body.
Lifelong experience has taught us that, in general, instances of large classes are
recalledbetterandfasterthaninstancesoflessfrequentclasses;thatlikelyoccurrencesare
easiertoimaginethanunlikelyones;andthattheassociativeconnectionsbetweenevents
arestrengthenedwhentheeventsfrequentlyco-occur.Asaresult,manhasathisdisposal
a procedure (the availability heuristic) for estimating the numerosity of a class, the
likelihood of an event, or the frequency of co-occurrences, by the ease with which the
relevant mental operations of retrieval, construction, or association can be performed.
However, as the preceding examples have demonstrated, this valuable estimation
procedureresultsinsystematicerrors.
AdjustmentandAnchoring
Inmanysituations,peoplemakeestimatesbystartingfromaninitialvaluethatisadjusted
to yield the final answer. The initial value, or starting point, may be suggested by the
formulationoftheproblem,oritmaybetheresultofapartialcomputation.Ineithercase,
adjustments are typically insufficient.18 That is, different starting points yield different
estimates,whicharebiasedtowardtheinitialvalues.Wecallthisphenomenonanchoring.
Insufficient adjustment. In a demonstration of the anchoring effect, subjects were
askedtoestimatevariousquantities,statedinpercentages(forexample,thepercentageof
AfricancountriesintheUnitedNations).Foreachquantity,anumberbetween0and100
was determined by spinning a wheel of fortune in the subjects’ presence. The subjects
wereinstructedtoindicatefirstwhetherthatnumberwashigherorlowerthanthevalueof
the quantity, and then to estimate the value of the quantity by moving upward or
downwardfromthegivennumber.Differentgroupsweregivendifferentnumbersforeach
quantity,andthesearbitrarynumbershadamarkedeffectonestimates.Forexample,the
medianestimatesofthepercentageofAfricancountriesintheUnitedNationswere25and
45forgroupsthatreceived10and65,respectively,asstartingpoints.Payoffsforaccuracy
didnotreducetheanchoringeffect.
Anchoringoccurs not only when thestarting point is given tothe subject, but also
whenthesubjectbaseshisestimateontheresultofsomeincompletecomputation.Astudy
ofintuitivenumericalestimationillustratesthiseffect.Twogroupsofhighschoolstudent
[choult os estimated, within 5 seconds, a numerical expression that was written on the
blackboard.Onegroupestimatedtheproduct
8×7×6×5×4×3×2×1
whileanothergroupestimatedtheproduct
1×2×3×4×5×6×7×8
To rapidly answer such questions, people may perform a few steps of computation and
estimate the product by extrapolation or adjustment. Because adjustments are typically
insufficient, this procedure should lead to underestimation. Furthermore, because the
resultofthefirstfewstepsofmultiplication(performedfromlefttoright)ishigherinthe
descending sequence than in the ascending sequence, the former expression should be
judgedlargerthan the latter. Bothpredictionswere confirmed. Themedianestimatefor
theascendingsequencewas512,whilethemedianestimateforthedescendingsequence
was2,250.Thecorrectansweris40,320.
Biases in the evaluation of conjunctive and disjunctive events.In a recentstudyby
Bar-Hillel19subjectsweregiventheopportunitytobetononeoftwoevents.Threetypes
ofeventswereused:(i)simpleevents,suchasdrawingaredmarblefromabagcontaining
50%redmarblesand50%whitemarbles;(ii)conjunctiveevents,suchasdrawingared
marble seven times in succession, with replacement, from a bag containing 90% red
marblesand10%whitemarbles;and(iii)disjunctiveevents,suchasdrawingaredmarble
atleastonceinsevensuccessivetries,withreplacement,fromabagcontaining10%red
marbles and 9% white marbles. In this problem, a significant majority of subjects
preferredtobetontheconjunctiveevent(theprobabilityofwhichis.48)ratherthanon
the simple event (the probability of which is .50). Subjects also preferred to bet on the
simpleevent rather thanon thedisjunctive event,which has aprobability of.52. Thus,
most subjects bet on the less likely event in both comparisons. This pattern of choices
illustrates a general finding. Studies of choice among gambles and of judgments of
probability indicate that people tend to overestimate the probability of conjunctive
events20 and to underestimate the probability of disjunctive events. These biases are
readilyexplainedaseffectsofanchoring.Thestatedprobabilityoftheelementaryevent
(success at any one stage) provides a natural starting point for the estimation of the
probabilities of both conjunctive and disjunctive events. Since adjustment from the
starting point is typically insufficient, the final estimates remain too close to the
probabilitiesoftheelementaryeventsinbothcases.Notethattheoverallprobabilityofa
conjunctive event is lower than the probability of each elementary event, whereas the
overallprobabilityofadisjunctiveeventishigherthantheprobabilityofeachelementary
event. As a consequence of anchoring, the overall probability will be overestimated in
conjunctiveproblemsandunderestimatedindisjunctiveproblems.
Biasesintheevaluationofcompoundeventsareparticularlysignificantinthecontext
ofplanning.Thesuccessfulcompletionofanundertaking,suchasthedevelopmentofa
newproduct,typicallyhasaconjunctivecharacter:fortheundertakingtosucceed,eachof
aseriesofeventsmustoccur.Evenwheneachoftheseeventsisverylikely,theoverall
probability of success can be quite low if the number of events is large. The general
tendency to overestimate the pr [timrall obability of conjunctive events leads to
unwarrantedoptimismintheevaluationofthelikelihoodthataplanwillsucceedorthata
project will be completed on time. Conversely, disjunctive structures are typically
encounteredintheevaluationofrisks.Acomplexsystem,suchasanuclearreactorora
human body, will malfunction if any of its essential components fails. Even when the
likelihoodoffailureineachcomponentisslight,theprobabilityofanoverallfailurecan
be high if many components are involved. Because of anchoring, people will tend to
underestimatetheprobabilitiesoffailureincomplexsystems.Thus,thedirectionofthe
anchoringbiascansometimesbeinferredfromthestructureoftheevent.Thechain-like
structureofconjunctionsleadstooverestimation,thefunnel-likestructureofdisjunctions
leadstounderestimation.
Anchoring in the assessment of subjective probability distributions. In decision
analysis,experts areoftenrequiredtoexpresstheirbeliefsaboutaquantity,suchasthe
value of the Dow Jones average on a particular day, in the form of a probability
distribution. Such a distribution is usually constructed by asking the person to select
valuesofthequantitythatcorrespondtospecifiedpercentilesofhissubjectiveprobability
distribution.Forexample,thejudgemaybeaskedtoselectanumber,X90,suchthathis
subjective probability that this number will be higher than the value of the Dow Jones
averageis.90.Thatis,heshouldselectthevalueX90sothatheisjustwillingtoaccept9
to 1 odds that the Dow Jones average will not exceed it. A subjective probability
distributionforthevalueoftheDowJonesaveragecanbeconstructedfromseveralsuch
judgmentscorrespondingtodifferentpercentiles.
By collecting subjective probability distributions for many different quantities, it is
possible to test the judge for proper calibration. A judge is properly (or externally)
calibratedinasetofproblemsifexactly %ofthetruevaluesoftheassessedquantities
fallsbelowhisstatedvaluesofX.Forexample,thetruevaluesshouldfallbelowX01
for 1% of the quantities and above X99 for 1% of the quantities. Thus, the true values
shouldfallintheconfidenceintervalbetweenX01andX99on98%oftheproblems.
Several investigators21 have obtained probability distributions for many quantities
from a large number of judges. These distributions indicated large and systematic
departures from proper calibration. In most studies, the actual values of the assessed
quantitiesareeithersmallerthanX0lorgreaterthanX99forabout30%oftheproblems.
Thatis,thesubjectsstateoverlynarrowconfidenceintervalswhichreflectmorecertainty
thanisjustifiedbytheirknowledgeabouttheassessedquantities.Thisbiasiscommonto
naiveandtosophisticatedsubjects,anditisnoteliminatedbyintroducingproperscoring
rules,whichprovideincentivesforexternalcalibration.Thiseffectisattributable,inpart
atleast,toanchoring.
To select X90 for the value of the Dow Jones average, for example, it is natural to
begin by thinking about one’s best estimate of the Dow Jones and to adjust this value
upward. If this adjustment—like most others—is insufficient, then X90 will not be
sufficientlyextreme.Asimilaranchoring[laricientlyeffectwilloccurintheselectionof
X10, which is presumably obtained by adjusting one’s best estimate downward.
Consequently, the confidence interval between X10and X90 will be too narrow, and the
assessedprobabilitydistributionwillbetootight.Insupportofthisinterpretationitcanbe
shown that subjective probabilities are systematically altered by a procedure in which
one’sbestestimatedoesnotserveasananchor.
Subjectiveprobabilitydistributionsforagivenquantity(theDowJonesaverage)can
beobtainedintwodifferentways:(i)byaskingthesubjecttoselectvaluesoftheDow
Jones that correspond to specified percentiles of his probability distribution and (ii) by
asking the subject to assess the probabilities that the true value of the Dow Jones will
exceed some specified values. The two procedures are formally equivalent and should
yield identical distributions. However, they suggest different modes of adjustment from
differentanchors.Inprocedure(i),thenaturalstartingpointisone’sbestestimateofthe
quantity.Inprocedure(ii),ontheotherhand,thesubjectmaybeanchoredonthevalue
statedinthequestion.Alternatively,hemaybeanchoredonevenodds,ora50–50chance,
whichisanaturalstartingpointintheestimationoflikelihood.Ineithercase,procedure
(ii)shouldyieldlessextremeoddsthanprocedure(i).
Tocontrastthetwoprocedures,asetof24quantities(suchastheairdistancefrom
NewDelhitoPeking)waspresentedtoagroupofsubjectswhoassessedeitherX10orX90
for each problem. Another group of subjects received the median judgment of the first
groupforeachofthe24quantities.Theywereaskedtoassesstheoddsthateachofthe
givenvaluesexceededthetruevalueoftherelevantquantity.Intheabsenceofanybias,
the second group should retrieve the odds specified to the first group, that is, 9:1.
However,ifevenoddsorthestatedvalueserveasanchors,theoddsofthesecondgroup
shouldbelessextreme,thatis,closerto1:1.Indeed,themedianoddsstatedbythisgroup,
across all problems, were 3:1. When the judgments of the two groups were tested for
external calibration, it was found that subjects in the first group were too extreme, in
accord with earlier studies. The events that they defined as having a probability of .10
actuallyobtainedin24%ofthecases.Incontrast,subjectsinthesecondgroupweretoo
conservative. Events to which they assigned an average probability of .34 actually
obtainedin26%ofthecases.Theseresultsillustratethemannerinwhichthedegreeof
calibrationdependsontheprocedureofelicitation.
Discussion
This article has been concerned with cognitive biases that stem from the reliance on
judgmental heuristics. These biases are not attributable to motivational effects such as
wishfulthinkingorthedistortionofjudgmentsbypayoffsandpenalties.Indeed,severalof
thesevereerrorsofjudgmentreportedearlieroccurreddespitethefactthatsubjectswere
encouragedtobeaccurateandwererewardedforthecorrectanswers.22
Therelianceonheuristicsandtheprevalenceofbiasesarenotrestrictedtolaymen.
Experiencedresearchersarealsopronetothesamebiases—whentheythinkintuitively.
For example, the tendency to predict the outcome that best represents the data, with
insufficientregardforpriorprobability,hasbeenobservedin the intuitivejudgmentsof
individuals who have had extensive training in statistics. [ticor pri23 Although the
statistically sophisticated avoid elementary errors, such as the gambler’s fallacy, their
intuitive judgments are liable to similar fallacies in more intricate and less transparent
problems.
Itisnotsurprisingthatusefulheuristicssuchasrepresentativenessandavailabilityare
retained,eventhoughtheyoccasionallyleadtoerrorsinpredictionorestimation.Whatis
perhaps surprising is the failure of people to infer from lifelong experience such
fundamentalstatisticalrulesasregressiontowardthemean,ortheeffectofsamplesizeon
sampling variability. Although everyone is exposed, in the normal course of life, to
numerous examples from which these rules could have been induced, very few people
discovertheprinciplesofsamplingandregressionontheirown.Statisticalprinciplesare
not learned from everyday experience because the relevant instances are not coded
appropriately. For example, people do not discover that successive lines in a text differ
moreinaveragewordlengththandosuccessivepages,becausetheysimplydonotattend
to the average word length of individual lines or pages. Thus, people do not learn the
relationbetweensamplesizeandsamplingvariability,althoughthedataforsuchlearning
areabundant.
Thelackofanappropriatecodealsoexplainswhypeopleusuallydonotdetectthe
biases in their judgments of probability. A person could conceivably learn whether his
judgments are externally calibrated by keeping a tally of the proportion of events that
actuallyoccuramongthosetowhichheassignsthesameprobability.However,itisnot
naturaltogroupeventsbytheirjudgedprobability.Intheabsenceofsuchgroupingitis
impossibleforanindividualtodiscover,forexample,thatonly50%ofthepredictionsto
whichhehasassignedaprobabilityof.9orhigheractuallycametrue.
The empirical analysis of cognitive biases has implications for the theoretical and
applied role of judged probabilities. Modern decision theory24 regards subjective
probability as the quantified opinion of an idealized person. Specifically, the subjective
probabilityofagiveneventisdefinedbythesetofbetsaboutthiseventthatsuchaperson
iswillingtoaccept.Aninternallyconsistent,orcoherent,subjectiveprobabilitymeasure
canbederivedforanindividualifhischoicesamongbetssatisfycertainprinciples,thatis,
theaxiomsofthetheory.Thederivedprobabilityissubjectiveinthesensethatdifferent
individuals are allowed to have different probabilities for the same event. The major
contribution of this approach is that it provides a rigorous subjective interpretation of
probability that is applicable to unique events and is embedded in a general theory of
rationaldecision.
It should perhaps be noted that, while subjective probabilities can sometimes be
inferred from preferences among bets, they are normally not formed in this fashion. A
person bets on team A rather than on team B because he believes that team A is more
likelytowin;hedoesnotinferthisbelieffromhisbettingpreferences.Thus,inreality,
subjectiveprobabilitiesdeterminepreferencesamongbetsandarenotderivedfromthem,
asintheaxiomatictheoryofrationaldecision.25
Theinherentlysubjectivenatureofprobabilityhasledmanystudentstothebeliefthat
coherence,orinternalconsistency,istheonlyvalidcriterionbywhichjudgedprobabilities
shouldbeevaluated.Fromthestandpointoftheformaltheoryofsubjectiveprobability,
any set of internally consistent probability judgments is as good as any other. This
criterion is not entirely satisfactory [ saf sub, because an internally consistent set of
subjective probabilities can be incompatible with other beliefs held by the individual.
Consider a person whose subjective probabilities for all possible outcomes of a coin-
tossinggamereflectthegambler’sfallacy.Thatis,hisestimateoftheprobabilityoftails
onaparticulartossincreaseswiththenumberofconsecutiveheadsthatprecededthattoss.
Thejudgmentsofsuchapersoncouldbeinternallyconsistentandthereforeacceptableas
adequate subjective probabilities according to the criterion of the formal theory. These
probabilities,however,areincompatiblewiththegenerallyheldbeliefthatacoinhasno
memory and is therefore incapable of generating sequential dependencies. For judged
probabilities to be considered adequate, or rational, internal consistency is not enough.
Thejudgmentsmustbecompatiblewiththeentirewebofbeliefsheldbytheindividual.
Unfortunately,therecanbenosimpleformalprocedureforassessingthecompatibilityof
asetofprobabilityjudgmentswiththejudge’stotalsystemofbeliefs.Therationaljudge
willneverthelessstriveforcompatibility,eventhoughinternalconsistencyismoreeasily
achieved and assessed. In particular, he will attempt to make his probability judgments
compatiblewithhisknowledgeaboutthesubjectmatter,thelawsofprobability,andhis
ownjudgmentalheuristicsandbiases.
Summary
This article described three heuristics that are employed in making judgments under
uncertainty:(i)representativeness,whichisusuallyemployedwhenpeopleareaskedto
judge the probability that an object or event A belongs to class or process B; (ii)
availabilityofinstancesorscenarios,whichisoftenemployedwhenpeopleareaskedto
assess the frequency of a class or the plausibility of a particular development; and (iii)
adjustment from an anchor, which is usually employed in numerical prediction when a
relevantvalueisavailable.Theseheuristicsarehighlyeconomicalandusuallyeffective,
but they lead to systematic and predictable errors. A better understanding of these
heuristicsandofthebiasestowhichtheyleadcouldimprovejudgmentsanddecisionsin
situationsofuncertainty.
P
Notes
1.
D.KahnemanandA.Tversky,“OnthePsychologyofPrediction,”PsychologicalReview
80(1973):237–51.
2.
Ibid.
3.
Ibid.
4.
D. Kahneman and A. Tversky, “Subjective Probability: A Judgment of
Representativeness,”CognitivePsychology3(1972):430–54.
5.
Ibid.
6.
W.Edwards,“ConservatisminHumanInformationProcessing,”inFormalRepresentation
ofHumanJudgment,ed.B.Kleinmuntz(NewYork:Wiley,1968),17–52.
[t=“orm
7.
KahnemanandTversky,“SubjectiveProbability.”
8.
A. Tversky and D. Kahneman, “Belief in the Law of Small Numbers,” Psychological
Bulletin76(1971):105–10.
9.
KahnemanandTversky,“OnthePsychologyofPrediction.”
10.
Ibid.
11.
Ibid.
12.
Ibid.
13.
A. Tversky and D. Kahneman, “Availability: A Heuristic for Judging Frequency and
Probability,”CognitivePsychology5(1973):207–32.
14.
Ibid.
15.
R. C. Galbraith and B. J. Underwood, “Perceived Frequency of Concrete and Abstract
Words,”Memory&Cognition1(1973):56–60.
16.
TverskyandKahneman,“Availability.”
17.
L.J.ChapmanandJ.P.Chapman,“GenesisofPopularbutErroneousPsychodiagnostic
Observations,”JournalofAbnormalPsychology73(1967):193–204;L.J.Chapmanand
J.P.Chapman,“IllusoryCorrelationasanObstacletotheUseofValidPsychodiagnostic
Signs,”JournalofAbnormalPsychology74(1969):271–80.
18.
P.SlovicandS.Lichtenstein,“ComparisonofBayesianandRegressionApproachestothe
Study of Information Processing in Judgment,” Organizational Behavior & Human
Performance6(1971):649–744.
19.
M. Bar-Hillel, “On the Subjective Probability of Compound Events,” Organizational
Behavior&HumanPerformance9(1973):396–406.
20.
J. Cohen, E. I. Chesnick, and D. Haran, “A Confirmation of the Inertial-? Effect in
SequentialChoiceandDecision,”BritishJournalofPsychology63(1972):41–46.
21.
M.Alpe[spa
Acta Psychologica 35 (1971): 478–94; R. L. Winkler, “The Assessment of Prior
Distributionsin Bayesian Analysis,”Journal of the American Statistical Association62
(1967):776–800.
22.
KahnemanandTversky,“SubjectiveProbability”;TverskyandKahneman,“Availability.”
23.
Kahneman and Tversky, “On the Psychology of Prediction”; Tversky and Kahneman,
“BeliefintheLawofSmallNumbers.”
24.
AppendixB:Choices,Values,AndFrames*
DanielKahnemanandAmosTversky
ABSTRACT: We discuss the cognitive and the psychophysical determinants of choice in
riskyandrisklesscontexts.Thepsychophysicsofvalueinduceriskaversioninthedomain
of gains and risk seeking in the domain of losses. The psychophysics of chance induce
overweighting of sure things and of improbable events, relative to events of moderate
probability.Decisionproblemscanbedescribedorframedinmultiplewaysthatgiverise
to different preferences, contrary to the invariance criterion of rational choice. The
process of mental accounting, in which people organize the outcomes of transactions,
explains some anomalies of consumer behavior. In particular, the acceptability of an
option can depend on whether a negative outcome is evaluated as a cost or as an
uncompensated loss. The relation between decision values and experience values is
discussed.
Making decisions is like speaking prose—people do it all the time, knowingly or
unknowingly.Itishardlysurprising,then,thatthetopicofdecisionmakingissharedby
many disciplines, from mathematics and statistics, through economics and political
science, to sociology and psychology. The study of decisions addresses both normative
and descriptive questions. The normative analysis is concerned with the nature of
rationality and the logic of decision making. The descriptive analysis, in contrast, is
concernedwithpeople’sbeliefsandpreferencesastheyare,not as theyshouldbe.The
tensionbetweennormativeanddescriptiveconsiderationscharacterizesmuchofthestudy
ofjudgmentandchoice.
Analyses of decision making commonly distinguish risky and riskless choices. The
paradigmaticexampleofdecisionun^v>
RiskyChoice
Riskychoices,suchaswhetherornottotakeanumbrellaandwhetherornottogotowar,
aremadewithoutadvanceknowledgeoftheirconsequences.Becausetheconsequencesof
suchactionsdependonuncertaineventssuchastheweatherortheopponent’sresolve,the
choiceofanactmaybeconstruedastheacceptanceofagamblethatcanyieldvarious
outcomes with different probabilities. It is therefore natural that the study of decision
making under risk has focused on choices between simple gambles with monetary
outcomesandspecifiedprobabilities,inthehopethatthesesimpleproblemswillreveal
basicattitudestowardriskandvalue.
Weshallsketchanapproachtoriskychoicethatderivesmanyofitshypothesesfrom
apsychophysicalanalysisofresponsestomoneyandtoprobability.Thepsychophysical
approach to decision making can be traced to a remarkable essay that Daniel Bernoulli
published in 1738 (Bernoulli 1954) in which he attempted to explain why people are
generally averse to risk and why risk aversion decreases with increasing wealth. To
illustrate risk aversion and Bernoulli’s analysis, consider the choice between a prospect
that offers an 85% chance to win $1,000 (with a 15% chance to win nothing) and the
alternativeofreceiving$800forsure.Alargemajorityofpeoplepreferthesurethingover
thegamble,althoughthegamblehashigher(mathematical)expectation.Theexpectation
ofamonetarygambleisaweightedaverage,whereeachpossibleoutcomeisweightedby
its probability of occurrence. The expectation of the gamble in this example is .85 ×
$1,000+.15×$0=$850,whichexceedstheexpectationof$800associatedwiththesure
thing. The preference for the sure gain is an instance of risk aversion. In general, a
preferenceforasureoutcomeoveragamblethathashigherorequalexpectationiscalled
risk averse, and the rejection of a sure thing in favor of a gamble of lower or equal
expectationiscalledriskseeking.
Bernoullisuggestedthatpeopledonotevaluateprospectsbytheexpectationoftheir
monetary outcomes, but rather by the expectation of the subjective value of these
outcomes.Thesubjectivevalueofagambleisagainaweightedaverage,butnowitisthe
subjective value of each outcome that is weighted by its probability. To explain risk
aversionwithinthisframework,Bernoulliproposed thatsubjectivevalue, or utility,isa
concavefunctionofmoney.Insuchafunction,thedifferencebetweentheutilitiesof$200
and$100,forexample,isgreaterthantheutilitydifferencebetween$1,200and$1,100.It
followsfromconcavitythatthesubjectivevalueattachedtoagainof$800ismorethan
80%ofthevalueofagainof$1,000.Consequently,theconcavityoftheutilityfunction
entailsariskaversepreferenceforasuregainof$800overan80%chancetowin$1,000,
althoughthetwoprospectshavethesamemonetaryexpectation.
Itiscustomaryindecisionanalysistodescribetheoutcomesofdecisionsintermsof
totalwealth.Forexample,anoffertobet$20onthetossofafaircoinisrepresentedasa
choicebetweenanindividual’scurrentwealthWandanevenchancetomovetoW+$20
ortoWnindispan>–$20.Thisrepresentationappearspsychologicallyunrealistic:People
donotnormallythinkofrelativelysmalloutcomesintermsofstatesofwealthbutrather
intermsofgains,losses,andneutraloutcomes(suchasthemaintenanceofthestatusquo).
If the effective carriers of subjective value are changes of wealth rather than ultimate
states of wealth, as we propose, the psychophysical analysis of outcomes should be
appliedtogainsandlossesratherthantototalassets.Thisassumptionplaysacentralrole
in a treatment of risky choice that we called prospect theory (Kahneman and Tversky
1979).Introspectionaswellaspsychophysicalmeasurementssuggestthatsubjectivevalue
isaconcavefunctionofthesizeofagain.Thesamegeneralizationappliestolossesas
well.Thedifferenceinsubjectivevaluebetweenalossof$200andalossof$100appears
greater than the difference in subjective value between a loss of $1,200 and a loss of
$1,100.Whenthevaluefunctionsforgainsandforlossesarepiecedtogether,weobtain
anS-shapedfunctionofthetypedisplayedinFigure1.
Figure1.AHypotheticalValueFunction
ThevaluefunctionshowninFigure1is(a)definedongainsandlossesratherthanon
totalwealth,(b)concaveinthedomainofgainsandconvexinthedomainoflosses,and
(c)considerablysteeperforlossesthanforgains.Thelastproperty,whichwelabelloss
aversion,expressestheintuitionthatalossof$Xismoreaversivethanagainof$Xis
attractive.Lossaversionexplainspeople’sreluctancetobetonafaircoinforequalstakes:
The attractiveness of the possible gain is not nearly sufficient to compensate for the
aversiveness of the possible loss. For example, most respondents in a sample of
undergraduatesrefusedto stake$10onthetossofacoiniftheystoodtowinlessthan
$30.
The assumption of risk aversion has played a central role in economic theory.
However,justastheconcavityofthevalueofgainsentailsriskaversion,theconvexityof
thevalueoflossesentailsriskseeking.Indeed,riskseekinginlossesisarobusteffect,
particularly when the probabilities of loss are substantial. Consider, for example, a
situationinwhichanindividualisforcedtochoosebetweenan85%chancetolose$1,000
(witha15%chancetolosenothing)andasurelossof$800.Alargemajorityofpeople
express a preference for the gamble over the sure loss. This is a risk seeking choice
becausetheexpectationof thegamble(–$850)is inferiortothe expectation ofthesure
loss (–$800). Risk seeking in the domain of losses has been confirmed by several
investigators(FishburnandKochenberger1979;HersheyandSchoemaker1980;Payne,
Laughhunn,andCrum1980;Slovic,Fischhoff,andLichtenstein1982).Ithasalsobeen
observedwithnonmonetaryoutcomes,suchashoursofpain(ErakerandSox1981)and
lossofhumanlives(Fischhoff1983;Tversky1977;TverskyandKahneman1981).Isit
wrongtoberiskaverseinthedomainofgainsandriskseekinginthedomainoflosses?
These preferences conform to compelling intuitions about the subjective value of gains
and losses, and the presumption is that people should be entitled to their own values.
However, we shall see that an S-shaped value function has implications that are
normativelyunacceptable.
Toaddressthenormativeissueweturnfrompsychologytodecisiontheory.Modern
decision theory can be said to begin with the pioneering work of von Neumann and
Morgenstern(1947),wholaiddownseveralqualitativeprinciples,oraxioms,thatshouldg
ctha211;$850)overn thepreferencesofarationaldecisionmaker.Theiraxiomsincluded
transitivity(ifAispreferredtoBandBispreferredtoC,thenAispreferredtoC),and
substitution(ifAispreferredtoB,thenanevenchancetogetAorCispreferredtoan
evenchancetogetBorC),alongwithotherconditionsofamoretechnicalnature.The
normativeandthedescriptivestatusoftheaxiomsofrationalchoicehavebeenthesubject
of extensive discussions. In particular, there is convincing evidence that people do not
always obey the substitution axiom, and considerable disagreement exists about the
normative merit of this axiom (e.g., Allais and Hagen 1979). However, all analyses of
rational choice incorporate two principles: dominance and invariance. Dominance
demandsthatifprospectAisatleastasgoodasprospectBineveryrespectandbetter
thanBinatleastonerespect,thenAshouldbepreferredtoB.Invariancerequiresthatthe
preferenceorderbetweenprospectsshouldnotdependonthemannerinwhichtheyare
described. In particular, two versions of a choice problem that are recognized to be
equivalent when shown together should elicit the same preference even when shown
separately. We now show that the requirement of invariance, however elementary and
innocuousitmayseem,cannotgenerallybesatisfied.
FramingofOutcomes
Riskyprospectsarecharacterizedbytheirpossibleoutcomesandbytheprobabilitiesof
theseoutcomes.Thesameoption,however,canbeframedordescribedindifferentways
(TverskyandKahneman1981).Forexample,thepossibleoutcomesofagamblecanbe
framed either as gains and losses relative to the status quo or as asset positions that
incorporate initial wealth. Invariance requires that such changes in the description of
outcomesshouldnotalterthepreferenceorder.Thefollowingpairofproblemsillustrates
a violation of this requirement. The total number of respondents in each problem is
denotedbyN,andthepercentagewhochoseeachoptionisindicatedinparentheses.
Problem 1 (N = 152): Imagine that the U.S. is preparing for the outbreak of an
unusual Asian disease, which is expected to kill 600 people. Two alternative
programstocombatthediseasehavebeenproposed.Assumethattheexactscientific
estimatesoftheconsequencesoftheprogramsareasfollows:
IfProgramAisadopted,200peoplewillbesaved.(72%)
IfProgramBisadopted,thereisaone-thirdprobabilitythat600peoplewillbe
savedandatwo-thirdsprobabilitythatnopeoplewillbesaved.(28%)
Whichofthetwoprogramswouldyoufavor?
TheformulationofProblem1implicitlyadoptsasareferencepointastateofaffairs
inwhichthediseaseisallowedtotakeitstollof600lives.Theoutcomesoftheprograms
includethereferencestateandtwopossiblegains,measuredbythenumberoflivessaved.
As expected, preferences are risk averse: A clear majority of respondents prefer saving
200livesforsureoveragamblethatoffersaone-thirdchanceofsaving600lives.Now
consider another problem in which the same cover story is followed by a different
descriptionoftheprospectsassociatedwiththetwoprograms:
Problem2(N=155):
IfProgramCisadopted,400peoplewilldie.(22%)
IfProgramDisadopted,thereisaone-thirdprobabilitythatnobodywilldieanda
two-thirdsprobabilitythat600peoplewilldie.(78%)
ItiseasytoverifythatoptionsCandDinProblem2areundistinguishableinreal
terms from options A and B in Problem 1, respectively. The second version, however,
assumesa reference state in whichno onedies of thedisease. Thebest outcome is the
maintenanceofthisstateandthealternativesarelossesmeasuredbythenumberofpeople
thatwilldieof the disease.Peoplewhoevaluateoptions in thesetermsareexpectedto
showariskseekingpreferenceforthegamble(optionD)overthesurelossof400lives.
Indeed,thereismoreriskseekinginthesecondversionoftheproblemthanthereisrisk
aversioninthefirst.
The failure of invariance is both pervasive and robust. It is as common among
sophisticated respondents as among naive ones, and it is not eliminated even when the
samerespondents answerboth questions withina fewminutes. Respondents confronted
with their conflicting answers are typically puzzled. Even after rereading the problems,
theystillwishtoberiskaverseinthe“livessaved”version;theywishtoberiskseekingin
the“liveslost”version;andtheyalsowishtoobeyinvarianceandgiveconsistentanswers
inthetwoversions.Intheirstubbornappeal,framingeffectsresembleperceptualillusions
morethancomputationalerrors.
The following pair of problems elicits preferences that violate the dominance
requirementofrationalchoice.
Problem3(N=86):Choosebetween:
E.25%chancetowin$240and75%chancetolose$760(0%)
F.25%chancetowin$250and75%chancetolose$750(100%)
ItiseasytoseethatFdominatesE.Indeed,allrespondentschoseaccordingly.
Problem 4 (N = 150): Imagine that you face the following pair of concurrent
decisions.
Firstexaminebothdecisions,thenindicatetheoptionsyouprefer.
Decision(i)Choosebetween:
A.asuregainof$240(84%)
B.25%chancetogain$1,000and75%chancetogainnothing(16%)
Decision(ii)Choosebetween:
C.asurelossof$750(13%)
D.75%chancetolose$1,000and25%chancetolosenothing(87%)
As expected from the previous analysis, a large majority of subjects made a risk
aversechoiceforthesuregainoverthepositivegambleinthefirstdecision,andaneven
largermajorityofsubjectsmadeariskseekingchoiceforthegambleoverthesurelossin
theseconddecision.Infact,73%oftherespondentschoseAandDandonly3%choseB
andC.The same cdCcefpattern of resultswasobservedin amodifiedversionof the
problem,withreducedstakes,inwhichundergraduatesselectedgamblesthattheywould
actuallyplay.
BecausethesubjectsconsideredthetwodecisionsinProblem4simultaneously,they
expressed in effect a preference for A and D over B and C. The preferred conjunction,
however,isactuallydominatedbytherejectedone.Addingthesuregainof$240(option
A)tooptionDyieldsa25%chancetowin$240anda75%chancetolose$760.Thisis
precisely option E in Problem 3. Similarly, adding the sure loss of $750 (option C) to
option B yields a 25% chance to win $250 and a 75% chance to lose $750. This is
precisely option F in Problem 3. Thus, the susceptibility to framing and the S-shaped
valuefunctionproduceaviolationofdominanceinasetofconcurrentdecisions.
The moral of these results is disturbing: Invariance is normatively essential,
intuitively compelling, and psychologically unfeasible. Indeed, we conceive only two
ways of guaranteeing invariance. The first is to adopt a procedure that will transform
equivalent versions of any problem into the same canonical representation. This is the
rationale for the standard admonition to students of business, that they should consider
each decision problem in terms of total assets rather than in terms of gains or losses
(Schlaifer1959).Sucharepresentationwouldavoidtheviolationsofinvarianceillustrated
in the previous problems, but the advice is easier to give than to follow. Except in the
context of possible ruin, it is more natural to consider financial outcomes as gains and
losses rather than as states of wealth. Furthermore, a canonical representation of risky
prospectsrequiresacompoundingofalloutcomesofconcurrentdecisions(e.g.,Problem
4) that exceeds the capabilities of intuitive computation even in simple problems.
Achieving a canonical representation is even more difficult in other contexts such as
safety,health,orqualityoflife.Shouldweadvisepeopletoevaluatetheconsequenceofa
publichealthpolicy(e.g.,Problems1and2)intermsofoverallmortality,mortalitydueto
diseases,orthenumberofdeathsassociatedwiththeparticulardiseaseunderstudy?
Anotherapproachthatcouldguaranteeinvarianceistheevaluationofoptionsinterms
oftheiractuarialratherthantheirpsychologicalconsequences.Theactuarialcriterionhas
some appeal in the context of human lives, but it is clearly inadequate for financial
choices, as has been generally recognized at least since Bernoulli, and it is entirely
inapplicabletooutcomesthatlackanobjectivemetric.Weconcludethatframeinvariance
cannotbeexpectedtoholdandthatasenseofconfidenceinaparticularchoicedoesnot
ensurethatthesamechoicewouldbemadeinanotherframe.Itisthereforegoodpractice
totesttherobustnessofpreferencesbydeliberateattemptstoframeadecisionproblemin
morethanoneway(Fischhoff,Slovic,andLichtenstein1980).
ThePsychophysicsofChances
OurdiscussionsofarhasassumedaBernoullianexpectationruleaccordingtowhichthe
value,orutility,ofanuncertainprospectisobtainedbyaddingtheutilitiesofthepossible
outcomes, each weighted by its probability. To examine this assumption, let us again
consultpsychophysicalintuitions. Setting thevalue of thestatus quo atzero, imagine a
cashgift,sayof$300,andassignitavalueofone.Nowimaginethatyouareonlygivena
tickettoalotterythathasasingleprizeof$300.Howdoesthevalueoftheticketvaryasa
functionoftheprobabilityofwinningtheprize?Barringutilityforgambling,thevalueof
suchaprospectmustvarybetweenzero(whenthechanceofwinningisnilcinntric.We)
andone(whenwinning$300isacertainty).
Intuitionsuggeststhatthevalueoftheticketisnotalinearfunctionoftheprobability
ofwinning,asentailedbytheexpectationrule.Inparticular,anincreasefrom0%to5%
appears to have a larger effect than an increase from 30% to 35%, which also appears
smaller than an increase from 95% to 100%. These considerations suggest a category-
boundaryeffect:Achangefromimpossibilitytopossibilityorfrompossibilitytocertainty
hasabiggerimpactthanacomparablechangeinthemiddleofthescale.Thishypothesis
isincorporatedintothecurvedisplayedinFigure2,whichplotstheweightattachedtoan
eventasafunctionofitsstatednumericalprobability.ThemostsalientfeatureofFigure2
isthatdecisionweightsareregressivewithrespecttostatedprobabilities.Exceptnearthe
endpoints, an increase of .05 in the probability of winning increases the value of the
prospectbylessthan5%ofthevalueoftheprize.Wenextinvestigatetheimplicationsof
thesepsychophysicalhypothesesforpreferencesamongriskyoptions.
Figure2.AHypotheticalWeightingFunction
InFigure2,decisionweightsarelowerthanthecorrespondingprobabilitiesovermost
of the range. Underweighting of moderate and high probabilities relative to sure things
contributes to risk aversion in gains by reducing the attractiveness of positive gambles.
Thesameeffectalsocontributestoriskseekinginlossesbyattenuatingtheaversiveness
of negative gambles. Low probabilities, however, are overweighted, and very low
probabilities are either overweighted quite grossly or neglected altogether, making the
decision weights highly unstable in that region. The overweightingof low probabilities
reversesthepatterndescribedabove:Itenhancesthevalueoflongshotsandamplifiesthe
aversiveness of a small chance of a severe loss. Consequently, people are often risk
seekingindealingwithimprobablegainsandriskaverseindealingwithunlikelylosses.
Thus,thecharacteristicsofdecisionweightscontributetotheattractivenessofbothlottery
ticketsandinsurancepolicies.
Thenonlinearityofdecisionweightsinevitablyleadstoviolationsofinvariance,as
illustratedinthefollowingpairofproblems:
Problem5(N=85):Considerthefollowingtwo-stagegame.Inthefirststage,there
is a 75% chance to end the game without winning anything and a 25% chance to
moveintothesecondstage.Ifyoureachthesecondstageyouhaveachoicebetween:
A.asurewinof$30(74%)
B.80%chancetowin$45(26%)
Yourchoicemustbemadebeforethegamestarts,i.e.,beforetheoutcomeofthefirst
stageisknown.Pleaseindicatetheoptionyouprefer.
Problem6(N=81):Whichofthefollowingoptionsdoyouprefer?
C.25%chancetowin$30(42%)
D.20%chancetowin$45(58%)
BecausethereisonechanceitoceinfourtomoveintothesecondstageinProblem
5,prospectAoffersa.25probabilityofwinning$30,andprospectBoffers.25×.80=.20
probability of winning $45. Problems 5 and 6 are therefore identical in terms of
probabilities and outcomes. However, the preferences are not the same in the two
versions:AclearmajorityfavorsthehigherchancetowinthesmalleramountinProblem
5,whereasthemajoritygoestheotherwayinProblem6.Thisviolationofinvariancehas
beenconfirmedwithbothrealandhypotheticalmonetarypayoffs(thepresentresultsare
withrealmoney),withhumanlivesasoutcomes,andwithanonsequentialrepresentation
ofthechanceprocess.
Weattributethefailureofinvariancetotheinteractionoftwofactors:theframingof
probabilitiesandthenonlinearityofdecisionweights.Morespecifically,weproposethat
inProblem5peopleignorethefirstphase,whichyieldsthesameoutcomeregardlessof
thedecisionthatismade,andfocustheirattentiononwhathappensiftheydoreachthe
second stage of the game. In that case, of course, they face a sure gain if they choose
optionAandan80%chanceofwinningiftheyprefertogamble.Indeed,people’schoices
inthesequentialversionarepracticallyidenticaltothechoicestheymakebetweenasure
gain of $30 and an 85% chance to win $45. Because a sure thing is overweighted in
comparisonwitheventsofmoderateorhighprobability,theoptionthatmayleadtoagain
of$30ismoreattractiveinthesequentialversion.Wecallthisphenomenonthepseudo-
certaintyeffectbecauseaneventthatisactuallyuncertainisweightedasifitwerecertain.
Acloselyrelatedphenomenoncanbedemonstratedatthelowendoftheprobability
range. Suppose you are undecided whether or not to purchase earthquake insurance
becausethepremiumisquitehigh.Asyouhesitate,yourfriendlyinsuranceagentcomes
forthwithanalternativeoffer:“Forhalftheregularpremiumyoucanbefullycoveredif
thequakeoccurs onanodd dayofthe month.Thisis a gooddealbecause forhalfthe
price you are covered for more than half the days.” Why do most people find such
probabilistic insurance distinctly unattractive? Figure 2 suggests an answer. Starting
anywhere in the region of low probabilities, the impact on the decision weight of a
reductionofprobabilityfromptop/2isconsiderablysmallerthantheeffectofareduction
fromp/2to0.Reducingtheriskbyhalf,then,isnotworthhalfthepremium.
The aversion to probabilistic insurance is significant for three reasons. First, it
underminestheclassicalexplanationofinsuranceintermsofaconcaveutilityfunction.
Accordingtoexpectedutilitytheory,probabilisticinsuranceshouldbedefinitelypreferred
tonormalinsurancewhenthelatterisjustacceptable(seeKahnemanandTversky1979).
Second,probabilisticinsurancerepresentsmanyformsofprotectiveaction,suchashaving
amedicalcheckup, buying new tires, orinstallingaburglaralarmsystem.Suchactions
typicallyreduce theprobability ofsome hazardwithout eliminatingit altogether. Third,
theacceptabilityofinsurancecanbemanipulatedbytheframingofthecontingencies.An
insurancepolicythatcoversfirebutnotflood,forexample,couldbeevaluatedeitheras
fullprotectionagainstaspecificrisk(e.g.,fire),orasareductionintheoverallprobability
of property loss. Figure 2 suggests that people greatly undervalue a reduction in the
probabilityofahazardincomparisontothecompleteeliminationofthathazard.Hence,
insuranceshouldappearmoreattractivewhenitisframedastheeliminationofriskthan
when it is described as a reduction of risk. Indeed, Slovic, Fischhoff, and Lichtenstein
(1982) showed that a hypotheti ct arnative cal vaccine that reduces the probability of
contractingadiseasefrom20%to10%islessattractiveifitisdescribedaseffectivein
halfofthecasesthanifitispresentedasfullyeffectiveagainstoneoftwoexclusiveand
equallyprobablevirusstrainsthatproduceidenticalsymptoms.
FormulationEffects
Sofarwehavediscussedframingasatooltodemonstratefailuresofinvariance.Wenow
turnattentiontotheprocessesthatcontroltheframingofoutcomesandevents.Thepublic
healthproblemillustratesaformulationeffectinwhichachangeofwordingfrom“lives
saved” to “lives lost” induced a marked shift of preference from risk aversion to risk
seeking.Evidently,thesubjectsadoptedthedescriptionsoftheoutcomesasgiveninthe
questionandevaluatedtheoutcomesaccordinglyasgainsorlosses.Anotherformulation
effect was reported by McNeil, Pauker, Sox, and Tversky (1982). They found that
preferences of physicians and patients between hypothetical therapies for lung cancer
varied markedly when their probable outcomes were described in terms of mortality or
survival.Surgery,unlikeradiationtherapy,entailsariskofdeathduringtreatment.Asa
consequence, the surgery option was relatively less attractive when the statistics of
treatmentoutcomesweredescribedintermsofmortalityratherthanintermsofsurvival.
Aphysician,andperhapsapresidentialadvisoraswell,couldinfluencethedecision
made by the patient or by the President, without distorting or suppressing information,
merely by the framing of outcomes and contingencies. Formulation effects can occur
fortuitously, without anyone being aware of the impact of the frame on the ultimate
decision.Theycanalsobeexploiteddeliberatelytomanipulatetherelativeattractiveness
of options. For example, Thaler (1980) noted that lobbyists for the credit card industry
insisted that any price difference between cash and credit purchases be labeled a cash
discountratherthanacreditcardsurcharge.Thetwolabelsframethepricedifferenceasa
gainorasalossbyimplicitlydesignatingeitherthelowerorthehigherpriceasnormal.
Becauselossesloomlargerthangains,consumersarelesslikelytoacceptasurchargethan
toforgoadiscount.Asistobeexpected,attemptstoinfluenceframingarecommoninthe
marketplaceandinthepoliticalarena.
The evaluation of outcomes is susceptible to formulation effects because of the
nonlinearity of the value function and the tendency of people to evaluate options in
relationtothereferencepointthatissuggestedorimpliedbythestatementoftheproblem.
It is worthy of note that in other contexts people automatically transform equivalent
messagesintothe samerepresentation.Studiesoflanguagecomprehensionindicatethat
people quickly recode much of what they hear into an abstract representation that no
longerdistinguisheswhethertheideawasexpressedinanactiveorinapassiveformand
nolongerdiscriminateswhatwasactuallysaidfromwhatwasimplied,presupposed,or
implicated (Clark and Clark 1977). Unfortunately, the mental machinery that performs
theseoperationssilentlyandeffortlesslyisnotadequatetoperformthetaskofrecoding
the two versions of the public health problem or the mortality survival statistics into a
commonabstractform.
TransactionsandTrades
Ouranalysisofframingandofvaluecanbeextendedtochoicesbetweenmultiattribute
options,suchastheacceptabilityofatransactionoratrade.Weproposethat,inorderto
evaluateamultiattributeoption,apersonsetsupamencsetoptiotalaccountthatspecifies
the advantages and the disadvantages associated with the option, relative to a
multiattributereferencestate.Theoverallvalueofanoptionisgivenbythebalanceofits
advantages and its disadvantages in relation to the reference state. Thus, an option is
acceptable if the value of its advantages exceeds the value of its disadvantages. This
analysis assumes psychological—but not physical—separability of advantages and
disadvantages.Themodeldoesnotconstrainthemannerinwhichseparateattributesare
combinedtoformoverallmeasuresofadvantageandofdisadvantage,butitimposeson
thesemeasuresassumptionsofconcavityandoflossaversion.
Our analysis of mental accounting owes a large debt to the stimulating work of
Richard Thaler (1980, 1985), who showed the relevance of this process to consumer
behavior.Thefollowingproblem,basedonexamplesofSavage(1954)andThaler(1980),
introducessomeoftherulesthatgoverntheconstructionofmentalaccountsandillustrates
theextensionoftheconcavityofvaluetotheacceptabilityoftransactions.
Problem7:Imaginethatyouareabouttopurchaseajacketfor$125andacalculator
for$15.Thecalculatorsalesmaninformsyouthatthecalculatoryouwishtobuyis
on sale for $10 at the other branch of the store, located 20 minutes’ drive away.
Wouldyoumakeatriptotheotherstore?
This problem is concerned with the acceptability of an option that combines a
disadvantage of inconvenience with a financial advantage that can be framed as a
minimal, topical, or comprehensive account. The minimal account includes only the
differences between the two options and disregards the features that they share. In the
minimalaccount,theadvantageassociatedwithdrivingtotheotherstoreisframedasa
gainof$5.Atopicalaccountrelatestheconsequencesofpossiblechoicestoareference
levelthatisdeterminedbythecontextwithinwhichthedecisionarises.Inthepreceding
problem,therelevanttopicisthepurchaseofthecalculator,andthebenefitofthetripis
thereforeframedasareductionoftheprice,from$15to$10.Becausethepotentialsaving
isassociatedonlywiththecalculator,thepriceofthejacketisnotincludedinthetopical
account.Thepriceofthejacket, as wellasotherexpenses,couldwellbe included ina
morecomprehensiveaccountinwhichthesavingwouldbeevaluatedinrelationto,say,
monthlyexpenses.
Theformulationoftheprecedingproblemappearsneutralwithrespecttotheadoption
ofaminimal,topical,orcomprehensiveaccount.Wesuggest,however,thatpeoplewill
spontaneouslyframedecisionsintermsoftopicalaccountsthat,inthecontextofdecision
making, play a role analogous to that of “good forms” in perception and of basic-level
categoriesincognition.Topicalorganization,inconjunctionwiththeconcavityofvalue,
entailsthatthewillingnesstotraveltotheotherstoreforasavingof$5onacalculator
shouldbeinverselyrelatedtothepriceofthecalculatorandshouldbeindependentofthe
priceofthejacket.Totestthisprediction,weconstructedanotherversionoftheproblem
inwhichthepricesofthetwoitemswereinterchanged.Thepriceofthecalculatorwas
givenas$125inthefirststoreand$120intheotherbranch,andthepriceofthejacket
wassetat$15.Aspredicted,theproportionsofrespondentswhosaidtheywouldmakethe
tripdifferedsharplyinthetwoproblems.Theresultsshowedthat68%oftherespondents
(N=88)werewillingtodrivetotheotherbranchtosave$5ona$15calculator,butonly
29%of93respondentswerewillingtomakethesametriptosave$5ona$125calculator.
ThisfindingcThinchsupportsthenotionoftopicalorganizationofaccounts,sincethetwo
versionsareidenticalbothintermsofaminimalandacomprehensiveaccount.
The significance of topical accounts for consumer behavior is confirmed by the
observationthatthestandarddeviationofthepricesthatdifferentstoresinacityquotefor
thesameproductisroughlyproportionaltotheaveragepriceofthatproduct(Pratt,Wise,
and Zeckhauser 1979). Since the dispersion of prices is surely controlled by shoppers’
effortstofindthebestbuy,theseresultssuggestthatconsumershardlyexertmoreeffortto
save$15ona$150purchasethantosave$5ona$50purchase.
Thetopicalorganizationofmentalaccountsleadspeopletoevaluategainsandlosses
inrelativeratherthaninabsoluteterms,resultinginlargevariationsintherateatwhich
moneyisexchangedforotherthings,suchasthenumberofphonecallsmadetofinda
goodbuyorthewillingnesstodrivealongdistancetogetone.Mostconsumerswillfind
iteasiertobuyacarstereosystemoraPersianrug,respectively,inthecontextofbuyinga
carorahousethanseparately.Theseobservations,ofcourse,runcountertothestandard
rationaltheoryofconsumerbehavior,whichassumesinvarianceanddoesnotrecognize
theeffectsofmentalaccounting.
Thefollowingproblemsillustrateanotherexampleofmentalaccountinginwhichthe
postingofacosttoanaccountiscontrolledbytopicalorganization:
Problem 8 (N= 200): Imagine that you have decided to see a play and paid the
admission price of $10 per ticket. As you enter the theater, you discover that you
havelosttheticket.Theseatwasnotmarked,andtheticketcannotberecovered.
Wouldyoupay$10foranotherticket?
Yes(46%)No(54%)
Problem9(N=183):Imaginethatyouhavedecidedtoseeaplaywhereadmissionis
$10perticket.Asyouenterthetheater,youdiscoverthatyouhavelosta$10bill.
Wouldyoustillpay$10foraticketfortheplay?
Yes(88%)No(12%)
Thedifferencebetweentheresponsestothetwoproblemsisintriguing.Whyaresomany
peopleunwillingtospend$10afterhavinglostaticket,iftheywouldreadilyspendthat
sumafterlosinganequivalentamountofcash?Weattributethedifferencetothetopical
organizationofmentalaccounts.Goingtothetheaterisnormallyviewedasatransaction
inwhichthecostoftheticketisexchangedfortheexperienceofseeingtheplay.Buyinga
second ticket increases the cost of seeing the play to a level that many respondents
apparentlyfindunacceptable.Incontrast,thelossofthecashisnotpostedtotheaccount
of the play, and it affects the purchase of a ticket only by making the individual feel
slightlylessaffluent.
An interesting effect was observed when the two versions of the problem were
presented to the same subjects. The willingness to replace a lost ticket increased
significantlywhenthatproblemfollowedthelost-cashversion.Incontrast,thewillingness
to buy a ticket after losing cash was not affected by prior presentation of the other
problem. The juxtaposition of the two problems apparent clemosition ly enabled the
subjectstorealizethatitmakessensetothinkofthelostticketaslostcash,butnotvice
versa.
The normative status of the effects of mental accounting is questionable. Unlike
earlierexamples, suchas thepublic healthproblem, inwhich thetwo versionsdiffered
only in form, it can be argued that the alternative versions of the calculator and ticket
problemsdifferalsoinsubstance.Inparticular,itmaybemorepleasurabletosave$5ona
$15purchasethanonalargerpurchase,anditmaybemoreannoyingtopaytwiceforthe
sameticketthantolose$10incash.Regret,frustration,andself-satisfactioncanalsobe
affectedbyframing(KahnemanandTversky1982).Ifsuchsecondaryconsequencesare
considered legitimate, then the observed preferences do not violate the criterion of
invarianceandcannotreadilyberuledoutasinconsistentorerroneous.Ontheotherhand,
secondaryconsequencesmaychangeuponreflection.Thesatisfactionofsaving$5ona
$15 item can be marred if the consumer discovers that she would not have exerted the
sameefforttosave$10ona$200purchase.Wedonotwishtorecommendthatanytwo
decision problems that have the same primary consequences should be resolved in the
same way. We propose, however, that systematic examination of alternative framings
offersausefulreflectivedevicethatcanhelpdecisionmakersassessthevaluesthatshould
beattachedtotheprimaryandsecondaryconsequencesoftheirchoices.
LossesandCosts
Manydecisionproblemstaketheformofachoicebetweenretainingthestatusquoand
acceptinganalternativetoit,whichisadvantageousinsomerespectsanddisadvantageous
inothers.Theanalysisofvaluethatwasappliedearliertounidimensionalriskyprospects
canbeextendedtothiscasebyassumingthatthestatusquodefinesthereferencelevelfor
all attributes. The advantages of alternative options will then be evaluated as gains and
theirdisadvantagesaslosses.Becauselossesloomlargerthangains,thedecisionmaker
willbebiasedinfavorofretainingthestatusquo.
Thaler (1980) coined the term “endowment effect” to describe the reluctance of
peopletopartfromassetsthatbelongtotheirendowment.Whenitismorepainfultogive
upanassetthanitispleasurabletoobtainit,buyingpriceswillbesignificantlylowerthan
sellingprices.Thatis,thehighestpricethatanindividualwillpaytoacquireanassetwill
besmallerthantheminimalcompensationthatwouldinducethesameindividualtogive
upthatasset,onceacquired.Thalerdiscussedsomeexamplesoftheendowmenteffectin
the behavior of consumers and entrepreneurs. Several studies have reported substantial
discrepanciesbetweenbuyingandsellingpricesinbothhypotheticalandrealtransactions
(Gregory 1983; Hammack and Brown 1974; Knetsch and Sinden 1984). These results
have been presented as challenges to standard economic theory, in which buying and
sellingpricescoincideexceptfortransactioncostsandeffectsofwealth.Wealsoobserved
reluctancetotradeinastudyofchoicesbetweenhypotheticaljobsthatdifferedinweekly
salary(S) and in the temperature (T) of the workplace. Our respondents were asked to
imaginethattheyheldaparticularposition(S1,T1)andwereofferedtheoptionofmoving
toadifferentposition(S2,T2),whichwasbetterinonerespectandworseinanother.We
foundthatmostsubjectswhowereassignedto(S1,T1)didnotwishtomoveto(S2,T2),
andc2<thatmostsubjectswhowereassignedtothelatterpositiondidnotwishtomove
totheformer.Evidently,thesamedifferenceinpayorinworkingconditionsloomslarger
asadisadvantagethanasanadvantage.
In general, loss aversion favors stability over change. Imagine two hedonically
identicaltwinswhofindtwoalternativeenvironmentsequallyattractive.Imaginefurther
thatbyforceofcircumstancethetwinsareseparatedandplacedinthetwoenvironments.
Assoonastheyadopttheirnewstatesasreferencepointsandevaluatetheadvantagesand
disadvantages of each other’s environments accordingly, the twins will no longer be
indifferentbetweenthetwostates,andbothwillprefertostaywheretheyhappentobe.
Thus, the instability of preferences produces a preference for stability. In addition to
favoringstabilityoverchange,thecombinationofadaptationandlossaversionprovides
limited protection against regret and envy by reducing the attractiveness of foregone
alternativesandofothers’endowments.
Lossaversionandtheconsequentendowmenteffectareunlikelytoplayasignificant
role in routine economic exchanges. The owner of a store, for example, does not
experience money paid to suppliers as losses and money received from customers as
gains.Instead,themerchantaddscostsandrevenuesoversomeperiodoftimeandonly
evaluates the balance. Matching debits and credits are effectively canceled prior to
evaluation. Payments made by consumers are also not evaluated as losses but as
alternative purchases. In accord with standard economic analysis, money is naturally
viewedasaproxyforthegoodsandservicesthatitcouldbuy.Thismodeofevaluationis
madeexplicitwhenanindividualhasinmindaparticularalternative,suchas,“Icaneither
buy a new camera or a new tent.” In this analysis, a person will buy a camera if its
subjectivevalueexceedsthevalueofretainingthemoneyitwouldcost.
Therearecasesinwhichadisadvantagecanbeframedeitherasacostorasaloss.In
particular,thepurchaseofinsurancecanalsobeframedasachoicebetweenasureloss
andtheriskofagreaterloss.Insuchcasesthecost-lossdiscrepancycanleadtofailuresof
invariance. Consider, for example, the choice between a sure loss of $50 and a 25%
chancetolose$200.Slovic,Fischhoff,andLichtenstein(1982)reportedthat80%oftheir
subjectsexpressedarisk-seekingpreferenceforthegambleoverthesureloss.However,
only35%ofsubjectsrefusedtopay$50forinsuranceagainsta25%riskoflosing$200.
SimilarresultswerealsoreportedbySchoemakerandKunreuther(1979)andbyHershey
andSchoemaker(1980).Wesuggestthatthesameamountofmoneythatwasframedasan
uncompensated loss in the first problem was framed as the cost of protection in the
second.Themodalpreferencewasreversedinthetwoproblemsbecauselossesaremore
aversivethancosts.
We have observed a similar effect in the positive domain, as illustrated by the
followingpairofproblems:
Problem10:Wouldyouacceptagamblethatoffersa10%chancetowin$95anda
90%chancetolose$5?
Problem11:Wouldyoupay$5toparticipateinalotterythatoffersa10%chanceto
win$100anda90%chancetowinnothing?
A total of 132 undergraduates answered the two questions, which were separated by a
short filler problem. The order of the questions was reversed for half the respondents.
Although it is easily confirmed that the two problems offer objecti coffler problevely
identical options, 55 of the respondents expressed different preferences in the two
versions.Amongthem,42rejectedthegambleinProblem10butacceptedtheequivalent
lotteryin Problem11.Theeffectivenessofthisseeminglyinconsequentialmanipulation
illustratesboththecost-lossdiscrepancyandthepowerofframing.Thinkingofthe$5asa
paymentmakestheventuremoreacceptablethanthinkingofthesameamountasaloss.
Theprecedinganalysisimpliesthatanindividual’ssubjectivestatecanbeimproved
by framing negative outcomes as costs rather than as losses. The possibility of such
psychological manipulations may explain a paradoxical form of behavior that could be
labeledthedead-losseffect.Thaler(1980)discussedtheexampleofamanwhodevelops
tenniselbowsoonafterpayingthemembershipfeeinatennisclubandcontinuestoplay
inagonytoavoidwastinghisinvestment.Assumingthattheindividualwouldnotplayif
he had not paid the membership fee, the question arises: How can playing in agony
improvetheindividual’slot?Playinginpain,wesuggest,maintainstheevaluationofthe
membershipfeeasacost.Iftheindividualweretostopplaying,hewouldbeforcedto
recognizethefeeasadeadloss,whichmaybemoreaversivethanplayinginpain.
ConcludingRemarks
Theconceptsofutilityandvaluearecommonlyusedintwodistinctsenses:(a)experience
value,thedegreeofpleasureorpain,satisfactionoranguishintheactualexperienceofan
outcome;and(b)decisionvalue,thecontributionofananticipatedoutcometotheoverall
attractivenessoraversivenessofanoptioninachoice.Thedistinctionisrarelyexplicitin
decisiontheorybecauseitistacitlyassumedthatdecisionvaluesandexperiencevalues
coincide.Thisassumptionispartoftheconceptionofanidealizeddecisionmakerwhois
abletopredictfutureexperienceswithperfectaccuracyandevaluateoptionsaccordingly.
For ordinary decision makers, however, the correspondence of decision values between
experiencevaluesisfarfromperfect(March1978).Somefactorsthataffectexperience
arenoteasilyanticipated,andsomefactorsthataffectdecisionsdonothaveacomparable
impactontheexperienceofoutcomes.
In contrast to the large amount of research on decision making, there has been
relativelylittlesystematicexplorationofthepsychophysicsthatrelatehedonicexperience
toobjectivestates.Themostbasicproblemofhedonicpsychophysicsisthedetermination
ofthelevelofadaptationoraspirationthatseparatespositivefromnegativeoutcomes.The
hedonic reference point is largely determined by the objective status quo, but it is also
affected by expectations and social comparisons. An objective improvement can be
experienced as a loss, for example, when an employee receives a smaller raise than
everyoneelseintheoffice.Theexperienceofpleasureorpainassociatedwithachangeof
state is also critically dependent on the dynamics of hedonic adaptation. Brickman and
Campbell’s(1971) concept ofthehedonictreadmillsuggeststheradicalhypothesisthat
rapidadaptationwillcausetheeffectsofanyobjectiveimprovementtobeshort-lived.The
complexityandsubtletyofhedonicexperiencemakeitdifficultforthedecisionmakerto
anticipatetheactualexperiencethatoutcomeswillproduce.Manyapersonwhoordereda
mealwhenravenouslyhungryhasadmittedtoabigmistakewhenthefifthcoursearrived
onthetable.Thecommonmismatchofdecisionvaluesandexperiencevaluesintroduces
anadditionalelementofuncertaintyinmanydecisionproblems.
Theprevalenceofframingeffectsandviolationsofinvariancefurthercomplicatesthe
relati ces maker won between decision values and experience values. The framing of
outcomesofteninducesdecisionvaluesthathavenocounterpartinactualexperience.For
example,the framingof outcomes oftherapies forlung cancer interms of mortality or
survivalisunlikelytoaffectexperience,althoughitcanhaveapronouncedinfluenceon
choice. In other cases, however, the framing of decisions affects not only decision but
experienceaswell.Forexample,theframingofanexpenditureasanuncompensatedloss
orasthepriceofinsurancecanprobablyinfluencetheexperienceofthatoutcome.Insuch
cases, the evaluation of outcomes in the context of decisions not only anticipates
experiencebutalsomoldsit.
References
Allais,M.,andO.Hagen,eds.1979.ExpectedUtilityHypothesesandtheAllaisParadox.
Hingham,MA:D.Reidel.
Bernoulli,D.1954[1738].“ExpositionofaNewTheoryontheMeasurementofRisk.”
Econometrica22:23–36.
Brickman, P., and D. T. Campbell. 1971. “Hedonic Relativism and Planning the Good
Society.”InAdaptationLevelTheory:ASymposium,ed.M.H.Appley.NewYork:
AcademicPress,287–302.
Clark,H.H.,andE.V.Clark.1977.PsychologyandLanguage.NewYork:Harcourt.
Erakar,S.E.,andH.C.Sox.1981.“AssessmentofPatients’PreferencesforTherapeutic
Outcomes.”MedicalDecisionMaking1:29–39.
Fischhoff,B.1983.“PredictingFrames.”JournalofExperimentalPsychology:Learning,
MemoryandCognition9:103–16.
Fischhoff,B.,P.Slovic,andS.Lichtenstein.1980.“KnowingWhatYouWant:Measuring
Labile Values.” In Cognitive Processes in Choice and Decision Behavior, ed. T.
Wallsten.Hillsdale,NJ:Erlbaum,117–41.
Fishburn,P.C.,andG.A.Kochenberger.1979.“Two-PiecevonNeumann–Morgenstern
UtilityFunctions.”DecisionSciences10:503–18.
Gregory, R. 1983. “Measures of Consumer’s Surplus: Reasons for the Disparity in
ObservedValues.”Unpublishedmanuscript,KeeneStateCollege,Keene,NH.
Hammack,J.,andG.M.BrownJr.1974.WaterfowlandWetlands:TowardBioeconomic
Analysis.Baltimore:JohnsHopkinsUniversityPress.
Hershey,J.C.,andP.J.H.Schoemaker.1980.“RiskTakingandProblemContextinthe
DomainofLosses:AnExpected-UtilityAnalysis.”JournalofRiskandInsurance47:
111–32.
Kahneman,D.,andA.Tversky.1979.“ProspectTheory:AnAnalysisofDecisionunder
Risk.”Econometrica47:263–91.
———.1982.“TheSimulationHeuristic.”InJudgmentUnderUncertainty:Heuristics
and Biases, ed. D. Kahneman, P. Slovic, and A. Tver c, aistsky. New York:
CambridgeUniversityPress,201–208.
Knetsch, J., and J. Sinden. 1984. “Willingness to Pay and Compensation Demanded:
ExperimentalEvidenceofanUnexpectedDisparityinMeasuresofValue.”Quarterly
JournalofEconomics99:507–21.
March, J. G. 1978. “Bounded Rationality, Ambiguity, and the Engineering of Choice.”
BellJournalofEconomics9:587–608.
McNeil, B., S. Pauker, H. Sox Jr., and A. Tversky. 1982. “On the Elicitation of
Preferences for Alternative Therapies.” New England Journal of Medicine 306:
1259–62.
Payne, J. W., D. J. Laughhunn, and R. Crum. 1980. “Translation of Gambles and
AspirationLevelEffectsinRiskyChoiceBehavior.”ManagementScience26:1039–
60.
Pratt,J.W.,D.Wise,andR.Zeckhauser.1979.“PriceDifferencesinAlmostCompetitive
Markets.”QuarterlyJournalofEconomics93:189–211.
Savage,L.J.1954.TheFoundationofStatistics.NewYork:Wiley.
Schlaifer, R. 1959. Probability and Statistics for Business Decisions. New York:
McGraw-Hill.
Schoemaker,P.J.H.,andH.C.Kunreuther.1979.“AnExperimentalStudyofInsurance
Decisions.”JournalofRiskandInsurance46:603–18.
Slovic, P., B. Fischhoff, and S. Lichtenstein. 1982. “Response Mode, Framing, and
InformationProcessing Effects in Risk Assessment.” In New Directions for
Methodology of Social andBehavioral Science: Question Framing and Response
Consistency,ed.R.Hogarth.SanFrancisco:Jossey-Bass,21–36.
Thaler,R.1980.“TowardaPositiveTheoryofConsumerChoice.”JournalofEconomic
BehaviorandOrganization1:39–60.
———. 1985. “Using Mental Accounting in a Theory of Consumer Behavior.”
MarketingScience4:199–214.
Tversky, A. 1977. “On the Elicitation of Preferences: Descriptive and Prescriptive
Considerations.”InConflictingObjectivesinDecisions,ed.D.Bell,R.L. Kenney,
andH.Raiffa.NewYork:Wiley,209–22.
Tversky,A.,andD.Kahneman.1981.“TheFramingofDecisionsandthePsychologyof
Choice.”Science211:453–58.
vonNeumann,J.,andO.Morgenstern.1947.TheoryofGamesandEconomicBehavior,
2nded.Princeton:PrincetonUniversityPress.
P
AlsobyDanielKahneman
InternationalDifferencesinWell-Bf,aisan
(writtenwithEdDienerandJohnF.Helliwell)
HeuristicsandBiases:ThePsychologyofIntuitiveJudgment
(editedwithThomasGilovichandDaleGriffin)
Choices,Values,andFrames(editedwithAmosTversky)
Well-Being:TheFoundationsofHedonicPsychology
(editedwithEdwardDienerandNorbertSchwartz)
JudgmentUnderUncertainty:HeuristicsandBiases
(editedwithPaulSlovicandAmosTversky)
AttentionandEffort
P
Acknowledgments
Iamfortunatetohavemanyfriendsandnoshameaboutaskingforhelp.Everyoneofmy
friendshasbeenapproached,someofthemmanytimes,withrequestsforinformationor
editorialsuggestions.Iapologizefornotlistingthemall.Afewindividualsplayedamajor
roleinmakingthebookhappen.MythanksgofirsttoJasonZweig,whourgedmeintothe
projectandpatientlytriedtoworkwithmeuntilitbecamecleartobothofusthatIam
impossibletoworkwith.Throughout,hehasbeengenerouswithhiseditorialadviceand
enviable erudition, and sentences that he suggested dot the book. Roger Lewin turned
transcriptsofasetoflecturesintochapterdrafts.MaryHimmelsteinprovidedvaluable
assistancethroughout.JohnBrockmanbeganasanagentandbecameatrustedfriend.Ran
Hassinprovidedadviceandencouragementwhenitwasmostneeded.Inthefinalstages
ofalongjourneyIhadtheindispensablehelpofEricChinski,myeditoratFarrar,Straus
and Giroux. He knew the book better than I did and the work became an enjoyable
collaboration—I had not imagined that an editor could do as much as Eric did. My
daughter, Lenore Shoham, rallied round to help me through the hectic final months,
providingwisdom,asharpcriticaleye,andmanyofthesentencesinthe“Speakingof”
sections.Mywife,AnneTreisman,wentthroughalotanddidalot—Iwouldhavegiven
uplongagowithouthersteadysupport,wisdom,andendlesspatience.
P
Notes
Introduction
pronetocollecttoofewobservations:Wehadreadabookthatcriticizedpsychologistsfor
using small samples, but did not explain their choices: Jacob Cohen, Statistical Power
AnalysisfortheBehavioralSciences(Hillsdale,NJ:Erlbaum,1969).
questionaboutwords:Ihaveslightlyalteredtheoriginalwording,whichreferredtoletters
inthefirstandthirdpositionofwords.
negativeviewofthemind:AprominentGermanpsychologisthasbeenourmostpersistent
critic.GerdGigerenzer,“HowtoMakeCognitiveIllusionsDisappear,”EuropeanReview
ofSocialPsychology2(1991):83–115.GerdGigerenzer,“PersonalReflectionsonTheory
andPsychology,”Theory&Psychology20(2010):733–43.DanielKahnemanandAmos
Tversky,“OntheRealityofCognitiveIllusions,”PsychologicalReview103(1996):582–
91.
offeredplausiblealternatives:SomeexamplesfrommanyareValerieF.ReynaandFarrell
J. Lloyd, “Physician Decision-Making and Cardiac Risk: Effects of Knowledge, Risk
Perception,RiskToleranceandFuzzy-Processing,”JournalofExperimentalPsychology:
Applied12(2006):179–95.NicholasEpleyandThomasGilovich,“TheAnchoring-and-
AdjustmentHeuristic,”PsychologicalScience17(2006):311–18.NorbertSchwarzetal.,
“EaseofRetrievalofInformation:AnotherLookattheAvailabilityHeuristic,”Journalof
Personality and Social Psychology 61 (1991): 195–202. Elke U. Weber et al.,
“Asymmetric Discounting in Intertemporal Choice,” Psychological Science 18 (2007):
516–23. George F. Loewenstein et al., “Risk as Feelings,” Psychological Bulletin 127
(2001):267–86.
NobelPrizethatIreceived:TheprizeawardedineconomicsisnamedBankofSweden
PrizeinEconomicSciencesinMemoryofAlfredNobel.Itwasfirstgivenin1969.Some
physicalscientistswerenotpleasedwiththeadditionofaNobelPrizeinsocialscience,
andthedistinctivelabeloftheeconomicsprizewasacompromise.
prolongedpractice:HerbertSimonandhisstudentsatCarnegieMelloninthe1980sset
thefoundationsforourunderstandingofexpertise.Foranexcellentpopularintroduction
to the subject, see Joshua Foer, Moonwalking with Einstein: The Art and Science of
Remembering (New York: Penguin Press, 2011). He presents work that is reviewed in
more technical detail in K. Anders Ericsson et al., eds., The Cambridge Handbook of
ExpertiseandExpertPerformance(NewYork:CambridgeUniversityPress,2006.)
kitchenwasonfire:GaryA.Klein,SourcesofPower(Cambridge,MA:MITPress,1999).
studied chess masters: Herbert Simon was one of the great scholars of the twentieth
century,whosediscoveriesandinventionsrangedfrompoliticalscience(wherehebegan
hiscareer)toeconomics(inwhichhewonaNobelPrize)tocomputerscience(inwhich
hewasapioneer)andtopsychology.
“Thesituation…recognition”:HerbertA.Simon,“WhatIsanExplanationofBehavior?”
PsychologicalScience3(1992):150–61.
affect heuristic: The concept of the affect heuristic was developed by Paul Slovic, a
classmateofAmos’satMichiganandalifelongfriend.
withoutnoticingthesubstitution:.
1:TheCharactersoftheStory
offered many labels: For reviews of the field, see Jonathan St. B. T. Evans and Keith
Frankish,eds.,InTwoMinds:DualProcessesandBeyond(NewYork:OxfordUniversity
Press, 2009); Jonathan St. B. T. Evans, “Dual-Processing Accounts of Reasoning,
Judgment, and Social Cognition,” Annual Review of Psychology 59 (2008): 25 {59
eight=“0%“5–78. Among the pioneers are Seymour Epstein, Jonathan Evans, Steven
Sloman,KeithStanovich,andRichardWest.IborrowthetermsSystem1andSystem2
fromearlywritingsofStanovichandWestthatgreatlyinfluencedmythinking:KeithE.
StanovichandRichardF.West,“IndividualDifferencesinReasoning:Implicationsforthe
RationalityDebate,”BehavioralandBrainSciences23(2000):645–65.
subjectiveexperienceofagency:Thissenseoffreewillissometimesillusory,asshownin
Daniel M. Wegner, The Illusion of Conscious Will (Cambridge, MA: Bradford Books,
2003).
attentionis totally focusedelsewhere:NilliLavie,“Attention,DistractionandCognitive
ControlUnderLoad,”CurrentDirectionsinPsychologicalScience19(2010):143–48.
conflictbetweenthetwosystems:IntheclassicStrooptask,youareshownadisplayof
patchesofdifferentcolors,orofwordsprintedinvariouscolors.Yourtaskistocallout
the names of the colors, ignoring the words. The task is extremely difficult when the
coloredwordsarethemselvesnamesofcolor(e.g.,GREENprintedinred,followedbyY
ELLOWprintedingreen,etc.).
psychopathiccharm:ProfessorHarewrotemetosay,“Yourteacherwasright,”March16,
2011. Robert D. Hare, Without Conscience: The Disturbing World of the Psychopaths
AmongUs(NewYork:GuilfordPress,1999).PaulBabiakandRobertD.Hare,Snakesin
Suits:WhenPsychopathsGotoWork(NewYork:Harper,2007).
littlepeople:Agentswithinthemindarecalledhomunculiandare(quiteproperly)objects
ofprofessionalderision.
spaceinyourworkingmemory:AlanD.Baddeley,“WorkingMemory:LookingBackand
LookingForward,”NatureReviews:Neuroscience4(2003):829–38.AlanD.Baddeley,
YourMemory:AUser’sGuide(NewYork:FireflyBooks,2004).
2:AttentionandEffort
Attention and Effort: Much of the material of this chapter draws on my Attention and
Effort (1973). It is available for free download on my website
(www.princeton.edu/~kahneman/docs/attention_and_effort/Attention_hi_quality.pdf). The
mainthemeofthatbookistheideaofalimitedabilitytopayattentionandexertmental
effort. Attention and effort were considered general resources that could be used to
supportmanymentaltasks.Theideaofgeneralcapacityiscontroversial,butithasbeen
extended by other psychologists and neuroscientists, who found support for it in brain
research. See Marcel A. Just and Patricia A. Carpenter, “A Capacity Theory of
Comprehension: Individual Differences in Working Memory,” Psychological Review 99
(1992): 122–49; Marcel A. Just et al., “Neuroindices of Cognitive Workload:
Neuroimaging, Pupillometric and Event-Related Potential Studies of Brain Work,”
Theoretical Issues in Ergonomics Science 4 (2003): 56–88. There is also growing
experimentalevidenceforgeneral-purposeresourcesofattention,asinEvieVergauweet
al.,“DoMentalProcessesShareaDomain-GeneralResource?”PsychologicalScience21
(2010):384–90.Thereisimagingevidencethatthemereanticipationofahigh-efforttask
mobilizesactivityinmanyareasofthebrain,relativetoalow-efforttaskofthesamekind.
CarstenN.Boehler et al.,“Task-Load-Dependent Activation ofDopaminergicMidbrain
AreasintheAbsenceofReward,”JournalofNeuroscience31(2011):4955–61.
pupil of the eye: Eckhard H. Hess, “Attitude and Pupil Size,” Scientific American 212
(1965):46–54.
onthesubject’smind:Thewordsubjectremindssomepeopleofsubjugationandslavery,
and the American Psychological Association enjoins us to use the more democratic
participant. Unfortunately, the politically correct label is a mouthful, which occupies
memoryspaceandslowsthinking.Iwilldomybesttouseparticipantwheneverpossible
butwillswitchtosubjectwhennecessary.
heartrateincreases:DanielKahnemanetal.,“Pupillary,HeartRate,andSkinResistance
ChangesDuringaMentalTask,”JournalofExperimentalPsychology79(1969):164–67.
rapidlyflashingletters:DanielKahneman,JacksonBeatty,andIrwinPollack,“Perceptual
DeficitDuringaMentalTask,”Science15(1967):218–19.Weusedahalfwaymirrorso
thattheobserverssawthelettersdirectlyinfrontofthemwhilefacingthecamera.Ina
control condition, the participants looked at the letter through a narrow aperture, to
preventanyeffectofthechangingpupilsizeontheirvisualacuity.Theirdetectionresults
showedtheinverted-Vpatternobservedwithothersubjects.
Muchliketheelectricitymeter:Attemptingtoperformseveraltasksatoncemayruninto
difficultiesofseveralkinds.Forexample,itisphysicallyimpossibletosaytwodifferent
thingsatexactlythesametime,anditmaybeeasiertocombineanauditoryandavisual
taskthantocombinetwovisualortwoauditorytasks.Prominentpsychologicaltheories
have attempted to attribute all mutual interference between tasks to competition for
separate mechanisms. See Alan D. Baddeley, Working Memory (New York: Oxford
UniversityPress,1986).Withpractice,people’sabilitytomultitaskinspecificwaysmay
improve.However,thewidevarietyofverydifferenttasksthatinterferewitheachother
supportstheexistenceofageneralresourceofattentionoreffortthatisnecessaryinmany
tasks.
Studies of the brain: Michael E. Smith, Linda K. McEvoy, and Alan Gevins,
“NeurophysiologicalIndices of StrategyDevelopment andSkillAcquisition,” Cognitive
BrainResearch7(1999):389–404.AlanGevinsetal.,“High-ResolutionEEGMapping
of Cortical Activation Related to Working Memory: Effects of Task Difficulty, Type of
ProcessingandPractice,”CerebralCortex7(1997):374–85.
lessefforttosolvethesameproblems:Forexample,SylviaK.AhernandJacksonBeatty
showedthatindividualswhoscoredhigherontheSATshowedsmallerpupillarydilations
than low scorers in responding to the same task. “Physiological Signs of Information
ProcessingVarywithIntelligence,”Science205(1979):1289–92.
“law of least effort”: Wouter Kool et {ute979): 1289al., “Decision Making and the
Avoidance of Cognitive Demand,” Journal of Experimental Psychology—General 139
(2010): 665–82. Joseph T. McGuire and Matthew M. Botvinick, “The Impact of
Anticipated Demand on Attention and Behavioral Choice,” in Effortless Attention, ed.
BrianBruya(Cambridge,MA:BradfordBooks,2010),103–20.
balance of benefits and costs: Neuroscientists have identified a region of the brain that
assessestheoverallvalueofanactionwhenitiscompleted.Theeffortthatwasinvested
counts as a cost in this neural computation. Joseph T. McGuire and Matthew M.
Botvinick,“PrefrontalCortex,CognitiveControl,andtheRegistrationofDecisionCosts,”
PNAS107(2010):7922–26.
read distracting words: Bruno Laeng et al., “Pupillary Stroop Effects,” Cognitive
Processing12(2011):13–21.
associate with intelligence: Michael I. Posner and Mary K. Rothbart, “Research on
Attention Networks as a Model for the Integration of Psychological Science,” Annual
ReviewofPsychology58(2007):1–23.JohnDuncanetal.,“ANeuralBasisforGeneral
Intelligence,”Science289(2000):457–60.
undertimepressure:StephenMonsell,“TaskSwitching,”TrendsinCognitiveSciences7
(2003):134–40.
workingmemory:Baddeley,WorkingMemory.
testsofgeneralintelligence:AndrewA.Conway,MichaelJ.Kane,andRandallW.Engle,
“WorkingMemoryCapacityandItsRelationtoGeneralIntelligence,”TrendsinCognitive
Sciences7(2003):547–52.
Israeli Air Force pilots: Daniel Kahneman, Rachel Ben-Ishai, and Michael Lotan,
“RelationofaTestofAttentiontoRoadAccidents,”Journal of Applied Psychology 58
(1973):113–15.DanielGopher,“ASelectiveAttentionTestasaPredictorofSuccessin
FlightTraining,”HumanFactors24(1982):173–83.
3:TheLazyController
“optimal experience”: Mihaly Csikszentmihalyi, Flow: The Psychology of Optimal
Experience(NewYork:Harper,1990).
sweet tooth: Baba Shiv and Alexander Fedorikhin, “Heart and Mind in Conflict: The
InterplayofAffectandCognitioninConsumerDecisionMaking,”JournalofConsumer
Research 26 (1999): 278–92. Malte Friese, Wilhelm Hofmann, and Michaela Wänke,
“When Impulses Take Over: Moderated Predictive Validity of Implicit and Explicit
Attitude Measures in Predicting Food Choice and Consumption Behaviour,” British
JournalofSocialPsychology47(2008):397–419.
cognitively busy: Daniel T. Gilbert, “How Mental Systems Believe,” American
Psychologist 46 (1991): 107–19. C. Neil Macrae and Galen V. Bodenhausen, “Social
Cognition: Thinking Categorically about Others,” Annual Review of Psychology 51
(2000):93–120.
po{“><21;: Sian L. Beilockand ThomasH. Carr, “When High-PoweredPeople Fail:
Working Memory and Choking Under Pressure in Math,” Psychological Science 16
(2005):101–105.
exertionofself-control:MartinS.Haggeretal.,“EgoDepletionandtheStrengthModelof
Self-Control:AMeta-Analysis,”PsychologicalBulletin136(2010):495–525.
resisttheeffectsofegodepletion:MarkMuravenandElisavetaSlessareva,“Mechanisms
of Self-Control Failure: Motivation and Limited Resources,” Personality and Social
Psychology Bulletin 29 (2003): 894–906. Mark Muraven, Dianne M. Tice, and Roy F.
Baumeister,“Self-ControlasaLimitedResource:RegulatoryDepletionPatterns,”Journal
ofPersonalityandSocialPsychology74(1998):774–89.
morethanameremetaphor:MatthewT.Gailliotetal.,“Self-ControlReliesonGlucoseas
aLimitedEnergySource:WillpowerIsMoreThanaMetaphor,”JournalofPersonality
and Social Psychology 92 (2007): 325–36. Matthew T. Gailliot and Roy F. Baumeister,
“ThePhysiologyofWillpower:LinkingBloodGlucosetoSelf-Control,”Personalityand
SocialPsychologyReview11(2007):303–27.
egodepletion:Gailliot,“Self-ControlReliesonGlucoseasaLimitedEnergySource.”
depletioneffectsin judgment: Shai Danziger, Jonathan Levav, and Liora Avnaim-Pesso,
“ExtraneousFactorsinJudicialDecisions,”PNAS108(2011):6889–92.
intuitive—incorrect—answer: Shane Frederick, “Cognitive Reflection and Decision
Making,”JournalofEconomicPerspectives19(2005):25–42.
syllogism as valid: This systematic error is known as the belief bias. Evans, “Dual-
ProcessingAccountsofReasoning,Judgment,andSocialCognition.”
call them more rational: KeithE. Stanovich,Rationality and the Reflective Mind (New
York:OxfordUniversityPress,2011).
cruel dilemma: Walter Mischel and Ebbe B. Ebbesen, “Attention in Delay of
Gratification,”JournalofPersonalityandSocialPsychology16(1970):329–37.
“There were no toys…distress”: Inge-Marie Eigsti et al., “Predicting Cognitive Control
from Preschool to Late Adolescence and Young Adulthood,” Psychological Science 17
(2006):478–84.
higher scores on tests of intelligence: Mischel and Ebbesen, “Attention in Delay of
Gratification.” Walter Mischel, “Processes in Delay of Gratification,” in Advances in
Experimental Social Psychology, Vol. 7, ed. Leonard Berkowitz (San Diego, CA:
AcademicPress,1974),249–92.WalterMischel,YuichiShoda,andMonicaL.Rodriguez,
“Delay of Gratification in Children,” Science 244 (1989): 933–38. Eigsti, “Predicting
CognitiveControlfromPreschooltoLateAdolescence.”
improvementwasmaintained:M.RosarioRued{Rocencaetal.,“Training,Maturation,
andGeneticInfluencesontheDevelopmentofExecutiveAttention,”PNAS 102 (2005):
14931–36.
conventionalmeasuresofintelligence:MaggieE.Toplak,RichardF.West,andKeithE.
Stanovich,“TheCognitiveReflectionTestasaPredictor of Performanceon Heuristics-
and-BiasesTasks,”Memory&Cognition(inpress).
4:TheAssociativeMachine
Associative Machine: Carey K. Morewedge and Daniel Kahneman, “Associative
ProcessesinIntuitiveJudgment,”TrendsinCognitiveSciences14(2010):435–40.
beyondyourcontrol:Toavoidconfusion,Ididnotmentioninthetextthatthepupilalso
dilated. The pupil dilates both during emotional arousal and when arousal accompanies
intellectualeffort.
thinkwithyourbody:PaulaM.Niedenthal,“EmbodyingEmotion,”Science316(2007):
1002–1005.
WASH primes SOAP: The image is drawn from the working of a pump. The first few
draws on a pump do not bring up any liquid, but they enable subsequent draws to be
effective.
“finds he it yellow instantly”: John A. Bargh, Mark Chen, and Lara Burrows,
“Automaticity of Social Behavior: Direct Effects of Trait Construct and Stereotype
ActivationonAction,”JournalofPersonalityandSocialPsychology71(1996):230–44.
words related to old age: Thomas Mussweiler, “Doing Is for Thinking! Stereotype
ActivationbyStereotypicMovements,”PsychologicalScience17(2006):17–21.
The Far Side: Fritz Strack, Leonard L. Martin, and Sabine Stepper, “Inhibiting and
FacilitatingConditionsoftheHumanSmile:ANonobtrusiveTestoftheFacialFeedback
Hypothesis,”JournalofPersonalityandSocialPsychology54(1988):768–77.
upsettingpictures:UlfDimberg,MonikaThunberg,andSaraGrunedal,“FacialReactions
to Emotional Stimuli: Automatically Controlled Emotional Responses,” Cognition and
Emotion16(2002):449–71.
listen to messages: Gary L. Wells and Richard E. Petty, “The Effects of Overt Head
Movementson Persuasion:Compatibilityand IncompatibilityofResponses,” Basicand
AppliedSocialPsychology1(1980):219–30.
increasethefundingofschools:JonahBerger,MarcMeredith,andS.ChristianWheeler,
“Contextual Priming: Where People Vote Affects How They Vote,” PNAS 105 (2008):
8846–49.
Reminders of money: Kathleen D. Vohs, “The Psychological Consequences of Money,”
Science314(2006):1154–56.
appeal of authoritarian ideas: Jeff Greenberg et al., “Evidence for Terror Management
Theory II: The Effect of Mortality Salience on Reactions to Those Who Threaten or
BolstertheCulturalWorldview,”JournalofPersonalityandSocialPsychology{gy
“Lady Macbeth effect”: Chen-Bo Zhong and Katie Liljenquist, “Washing Away Your
Sins:ThreatenedMoralityandPhysicalCleansing,”Science313(2006):1451–52.
preferredmouthwashoversoap:SpikeLeeandNorbertSchwarz,“DirtyHandsandDirty
Mouths: Embodiment of the Moral-Purity Metaphor Is Specific to the Motor Modality
InvolvedinMoralTransgression,”PsychologicalScience21(2010):1423–25.
at a British university: Melissa Bateson, Daniel Nettle, and Gilbert Roberts, “Cues of
BeingWatchedEnhanceCooperationinaReal-WorldSetting,”BiologyLetters2(2006):
412–14.
introducedtothatstranger:TimothyWilson’sStrangerstoOurselves(Cambridge, MA:
BelknapPress,2002) presentsa conceptofan“adaptiveunconscious”thatissimilarto
System1.
5:CognitiveEase
“Easy”and“Strained”:Thetechnicaltermforcognitiveeaseisfluency.
diverse inputs and outputs: Adam L. Alter and Daniel M. Oppenheimer, “Uniting the
TribesofFluencytoFormaMetacognitiveNation,”PersonalityandSocialPsychology
Review13(2009):219–35.
“Becoming Famous Overnight”: Larry L. Jacoby, Colleen Kelley, Judith Brown, and
Jennifer Jasechko, “Becoming Famous Overnight: Limits on the Ability to Avoid
Unconscious Influences of the Past,” Journal of Personality and Social Psychology 56
(1989):326–38.
nicely stated the problem: Bruce W. A. Whittlesea, Larry L. Jacoby, and Krista Girard,
“Illusions of Immediate Memory: Evidence of an Attributional Basis for Feelings of
FamiliarityandPerceptualQuality,”Journal of Memory and Language29(1990):716–
32.
The impression of familiarity: Normally, when you meet a friend you can immediately
placeandnamehim;youoftenknowwhereyoumethimlast,whathewaswearing,and
whatyousaidtoeachother.Thefeelingoffamiliaritybecomesrelevantonlywhensuch
specificmemoriesarenotavailable.Itisafallback.Althoughitsreliabilityisimperfect,
the fallback is much better than nothing. It is the sense of familiarity that protects you
fromtheembarrassmentofbeing(andacting)astonishedwhenyouaregreetedasanold
friendbysomeonewhoonlylooksvaguelyfamiliar.
“body temperature of a chicken”: Ian Begg, Victoria Armour, and Thérèse Kerr, “On
Believing What We Remember,” Canadian Journal of Behavioural Science 17 (1985):
199–214.
low credibility: Daniel M. Oppenheimer, “Consequences of Erudite Vernacular Utilized
Irrespective of Necessity: Problems with Using Long Words Needlessly,” Applied
CognitivePsychology20(2006):139–56.
whentheyrhymed:MatthewS.McGloneandJessicaTofighbakhsh,“BirdsofaFeather
FlockConjointly(?):RhymeasReas{RhyPsychologicalScience11(2000):424–28.
fictitiousTurkishcompanies:AnujK.ShahandDanielM.Oppenheimer,“EasyDoesIt:
The Role of Fluency in Cue Weighting,” Judgment and Decision Making Journal 2
(2007):371–79.
engagedandanalyticmode:AdamL.Alter,DanielM.Oppenheimer,NicholasEpley,and
Rebecca Eyre, “Overcoming Intuition: Metacognitive Difficulty Activates Analytic
Reasoning,”JournalofExperimentalPsychology—General136(2007):569–76.
picturesofobjects:PiotrWinkielmanandJohnT.Cacioppo,“MindatEasePutsaSmile
ontheFace:PsychophysiologicalEvidenceThatProcessingFacilitationIncreasesPositive
Affect,”JournalofPersonalityandSocialPsychology81(2001):989–1000.
small advantage: Adam L. Alter and Daniel M. Oppenheimer, “Predicting Short-Term
StockFluctuationsbyUsingProcessingFluency,”PNAS103(2006).MichaelJ.Cooper,
OrlinDimitrov,andP.RaghavendraRau,“ARose.combyAnyOtherName,”Journalof
Finance56(2001):2371–88.
clunky labels: Pascal Pensa, “Nomen Est Omen: How Company Names Influence
Shortand Long-Run Stock Market Performance,” Social Science Research Network
WorkingPaper,September2006.
mereexposureeffect:RobertB.Zajonc,“AttitudinalEffectsofMereExposure,”Journal
ofPersonalityandSocialPsychology9(1968):1–27.
favoriteexperiments:RobertB.ZajoncandD.W.Rajecki,“ExposureandAffect:AField
Experiment,”PsychonomicScience17(1969):216–17.
never consciously sees: Jennifer L. Monahan, Sheila T. Murphy, and Robert B. Zajonc,
“Subliminal Mere Exposure: Specific, General, and Diffuse Effects,” Psychological
Science11(2000):462–66.
inhabitingtheshell:D.W.Rajecki,“EffectsofPrenatalExposuretoAuditoryorVisual
StimulationonPostnatalDistressVocalizationsinChicks,”BehavioralBiology11(1974):
525–36.
“Theconsequences…socialstability”:RobertB.Zajonc,“MereExposure:AGatewayto
theSubliminal,”CurrentDirectionsinPsychologicalScience10(2001):227.
triadofwords:AnnetteBolte,ThomasGoschke,andJuliusKuhl,“EmotionandIntuition:
EffectsofPositiveandNegativeMoodonImplicitJudgments of SemanticCoherence,”
PsychologicalScience14(2003):416–21.
association is retrieved: The analysis excludes all cases in which the subject actually
foundthecorrectsolution.Itshowsthatevensubjectswhowillultimatelyfailtofinda
commonassociationhavesomeideaofwhetherthereisonetobefound.
increase cognitive ease: Sascha Topolinski and Fritz Strack, “The Architecture of
Intuition: Fluency and Affect Determine {ectition Intuitive Judgments of Semantic and
Visual Coherence and Judgments of Grammaticality in Artificial Grammar Learning,”
JournalofExperimentalPsychology—General138(2009):39–63.
doubledaccuracy:Bolte,Goschke,andKuhl,“EmotionandIntuition.”
formacluster:BarbaraFredrickson,Positivity:GroundbreakingResearchRevealsHowto
Embrace the Hidden Strength of Positive Emotions, Overcome Negativity, and Thrive
(NewYork:RandomHouse,2009).JosephP.ForgasandRebekahEast,“OnBeingHappy
and Gullible: Mood Effects on Skepticism and the Detection of Deception,” Journalof
ExperimentalSocialPsychology44(2008):1362–67.
smiling reaction: Sascha Topolinski et al., “The Face of Fluency: Semantic Coherence
Automatically Elicits a Specific Pattern of Facial Muscle Reactions,” Cognition and
Emotion23(2009):260–71.
“previousresearch…individuals”:SaschaTopolinskiandFritz Strack, “The Analysisof
Intuition:ProcessingFluencyandAffectinJudgmentsofSemanticCoherence,”Cognition
andEmotion23(2009):1465–1503.
6:Norms,Surprises,andCauses
Anobserver:DanielKahnemanandDaleT.Miller,“NormTheory:ComparingRealityto
ItsAlternatives,”PsychologicalReview93(1986):136–53.
“tattooonmyback”:JosJ.A.VanBerkum,“UnderstandingSentencesinContext:What
BrainWavesCanTellUs,”CurrentDirectionsinPsychologicalScience17(2008):376–
80.
thewordpickpocket:RanR.Hassin,JohnA.Bargh,andJamesS.Uleman,“Spontaneous
CausalInferences,”JournalofExperimentalSocialPsychology38(2002):515–22.
indicatesurprise:AlbertMichotte,ThePerceptionofCausality(Andover,MA:Methuen,
1963). Alan M. Leslie and Stephanie Keeble, “Do Six-Month-Old Infants Perceive
Causality?”Cognition25(1987):265–88.
explosive finale: Fritz Heider and Mary-Ann Simmel, “An Experimental Study of
ApparentBehavior,”AmericanJournalofPsychology13(1944):243–59.
identify bullies and victims: Leslie and Keeble, “Do Six-Month-Old Infants Perceive
Causality?”
aswedie:PaulBloom,“IsGodanAccident?”Atlantic,December2005.
7:AMachineforJumpingtoConclusions
elegant experiment: Daniel T. Gilbert, Douglas S. Krull, and Patrick S. Malone,
“Unbelieving the Unbelievable: Some Problems in the Rejection of False Information,”
JournalofPersonalityandSocialPsychology59(1990):601–13.
descriptionsoftwopeople:SolomonE.Asch,“Forming{#823.
ImpressionsofPersonality,”JournalofAbnormalandSocialPsychology41(1946):258–
90.
allsixadjectives:Ibid.
Wisdom of Crowds: James Surowiecki, The Wisdom of Crowds (New York: Anchor
Books,2005).
one-sided evidence: Lyle A. Brenner, Derek J. Koehler, and Amos Tversky, “On the
Evaluation of One-Sided Evidence,” Journal of Behavioral Decision Making 9 (1996):
59–70.
8:HowJudgmentsHappen
biological roots: Alexander Todorov, Sean G. Baron, and Nikolaas N. Oosterhof,
“Evaluating Face Trustworthiness: A Model-Based Approach,” Social Cognitive and
AffectiveNeuroscience3(2008):119–27.
friendlyorhostile:AlexanderTodorov,ChrisP.Said,AndrewD.Engell,andNikolaasN.
Oosterhof, “Understanding Evaluation of Faces on Social Dimensions,” Trends in
CognitiveSciences12(2008):455–60.
may spell trouble: Alexander Todorov, Manish Pakrashi, and Nikolaas N. Oosterhof,
“EvaluatingFacesonTrustworthinessAfterMinimalTimeExposure,”Social Cognition
27(2009):813–33.
Australia, Germany, and Mexico: Alexander Todorov et al., “Inference of Competence
fromFacesPredictElectionOutcomes,”Science308(2005):1623–26.CharlesC.Ballew
andAlexanderTodorov,“PredictingPoliticalElectionsfromRapidandUnreflectiveFace
Judgments,”PNAS104(2007):17948–53.ChristopherY.OlivolaandAlexanderTodorov,
“Electedin100Milliseconds:Appearance-BasedTraitInferencesandVoting,”Journalof
NonverbalBehavior34(2010):83–110.
watchlesstelevision:GabrielLenzandChappellLawson,“LookingthePart:Television
Leads Less Informed Citizens to Vote Based on Candidates’ Appearance,” American
JournalofPoliticalScience(forthcoming).
absenceofaspecifictaskset:AmosTverskyandDanielKahneman,“ExtensionalVersus
Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment,” Psychological
Review90(1983):293–315.
ExxonValdez:WilliamH.Desvousgesetal.,“MeasuringNaturalResourceDamageswith
Contingent Valuation: Tests of Validity and Reliability,” in Contingent Valuation: A
CriticalAssessment,ed.JerryA.Hausman(Amsterdam:North-Holland,1993),91–159.
sense of injustice: Stanley S. Stevens, Psychophysics: Introduction to Its Perceptual,
Neural,andSocialProspect(NewYork:Wiley,1975).
detected that the words rhymed: Mark S. Seidenberg and Michael K. Tanenhaus,
“Orthographic Effects on Rhyme Monitoring,” Journal of Experimental Psychology—
HumanLearningandMemory5(1979):546–54.
95–96sentencewasliterallytrue:SamGlucksberg,PatriciaGildea,andHowardG.Boo
{How>
JournalofVerbalLearningandVerbalBehavior21(1982):85–98.
9:AnsweringanEasierQuestion
an intuitive answer to it came readily to mind: An alternative approach to judgment
heuristicshasbeenproposedbyGerdGigerenzer,PeterM.Todd,andtheABCResearch
Group,inSimple Heuristics That Make Us Smart (New York: Oxford University Press,
1999).Theydescribe“fastandfrugal”formalproceduressuchas“Takethebest[cue],”
whichundersomecircumstancesgeneratequiteaccuratejudgmentsonthebasisoflittle
information. As Gigerenzer has emphasized, his heuristics are different from those that
AmosandIstudied,andhehasstressedtheiraccuracyratherthanthebiasestowhichthey
inevitablylead.Muchoftheresearchthatsupportsfastandfrugalheuristicusesstatistical
simulationstoshowthattheycouldworkinsomereal-lifesituations,buttheevidencefor
the psychological reality of these heuristics remains thin and contested. The most
memorablediscoveryassociatedwiththisapproachistherecognitionheuristic,illustrated
byanexamplethathasbecomewell-known:asubjectwhoisaskedwhichoftwocitiesis
largerandrecognizesoneofthemshouldguessthattheonesherecognizesislarger.The
recognitionheuristicworksfairlywellifthesubjectknowsthatthecitysherecognizesis
large; if she knows it to be small, however, she will quite reasonably guess that the
unknowncityislarger.Contrarytothetheory,thesubjectsusemorethantherecognition
cue: Daniel M. Oppenheimer, “Not So Fast! (and Not So Frugal!): Rethinking the
RecognitionHeuristic,”Cognition 90 (2003): B1–B9. A weakness of the theory is that,
fromwhatwe know of themind,thereisnoneedforheuristicstobefrugal.Thebrain
processesvastamountsofinformationinparallel,andthemindcanbefastandaccurate
without ignoring information. Furthermore, it has been known since the early days of
researchonchessmastersthatskillneednotconsistoflearningtouselessinformation.
Onthecontrary,skillismoreoftenanabilitytodealwithlargeamountsofinformation
quicklyandefficiently.
best examples of substitution: Fritz Strack, Leonard L. Martin, and Norbert Schwarz,
“PrimingandCommunication:SocialDeterminantsofInformationUseinJudgmentsof
LifeSatisfaction,”EuropeanJournalofSocialPsychology18(1988):429–42.
correlationsbetweenpsychologicalmeasures:Thecorrelationwas.66.
dominates happiness reports: Other substitution topics include marital satisfaction, job
satisfaction,andleisuretimesatisfaction:NorbertSchwarz,FritzStrack,andHans-Peter
Mai, “Assimilation and Contrast Effects in Part-Whole Question Sequences: A
ConversationalLogicAnalysis,”PublicOpinionQuarterly55(1991):3–23.
evaluatetheirhappiness:AtelephonesurveyconductedinGermanyincludedaquestion
aboutgeneralhappiness.Whentheself-reportsofhappinesswerecorrelatedwiththelocal
weatheratthetimeoftheinterview,apronouncedcorrelationwasfound.Moodisknown
to vary with the weather, and substitution explains the effect on reported happiness.
However, another version of the telephone survey yielded a somewhat different result.
These respondents were asked about the current weather before they were asked the
happiness quest {ppiournal ofion. For them, weather had no effect at all on reported
happiness! The explicit priming of weather provided them with an explanation of their
mood,underminingtheconnectionthatwouldnormallybemadebetweencurrentmood
andoverallhappiness.
view of the benefits: Melissa L. Finucane et al., “The Affect Heuristic in Judgments of
RisksandBenefits,”JournalofBehavioralDecisionMaking13(2000):1–17.
10:TheLawofSmallNumbers
“Itisboth…withoutadditives”:HowardWainerandHarrisL.Zwerling,“EvidenceThat
SmallerSchools DoNot Improve StudentAchievement,” Phi Delta Kappan88 (2006):
300–303.TheexamplewasdiscussedbyAndrewGelmanandDeborahNolan,Teaching
Statistics:ABagofTricks(NewYork:OxfordUniversityPress,2002).
50% risk of failing: Jacob Cohen, “The Statistical Power of Abnormal-Social
Psychological Research: A Review,” Journal of Abnormal and Social Psychology 65
(1962):145–53.
“BeliefintheLawofSmallNumbers”:AmosTverskyandDanielKahneman,“Beliefin
theLawofSmallNumbers,”PsychologicalBulletin76(1971):105–10.
“statisticalintuitions…wheneverpossible”:Thecontrastthatwedrewbetweenintuition
andcomputationseems to foreshadow thedistinctionbetweenSystems1and2,butwe
werealongwayfromtheperspectiveofthisbook.Weusedintuitiontocoveranything
butacomputation,anyinformalwaytoreachaconclusion.
German spies: William Feller, Introduction to Probability Theory and Its Applications
(NewYork:Wiley,1950).
randomness in basketball: Thomas Gilovich, Robert Vallone, and Amos Tversky, “The
Hot Hand in Basketball: On the Misperception of Random Sequences,” Cognitive
Psychology17(1985):295–314.
11:Anchors
“‘reasonable’ volume”: Robyn Le Boeuf and Eldar Shafir, “The Long and Short of It:
PhysicalAnchoringEffects,”JournalofBehavioralDecisionMaking19(2006):393–406.
nodtheirhead:NicholasEpleyandThomasGilovich,“PuttingAdjustmentBackinthe
Anchoring and Adjustment Heuristic: Differential Processing of Self-Generated and
Experimenter-ProvidedAnchors,”PsychologicalScience12(2001):391–96.
stay closer to the anchor: Epley and Gilovich, “The Anchoring-and-Adjustment
Heuristic.”
associative coherence: Thomas Mussweiler, “The Use of Category and Exemplar
Knowledge in the Solution of Anchoring Tasks,” Journal of Personality and Social
Psychology78(2000):1038–52.
San Francisco Exploratorium: Karen E. Jacowitz and Daniel Kahneman, “Measures of
AnchoringinEstimationTasks,”Person{pantionalityandSocialPsychologyBulletin21
(1995):1161–66.
substantiallylower:Gregory B.Northcraft and MargaretA. Neale,“Experts, Amateurs,
and Real Estate: An Anchoring-and-Adjustment Perspective on Property Pricing
Decisions,”Organizational Behavior and Human Decision Processes 39 (1987): 84–97.
The high anchor was 12% above the listed price, the low anchor was 12% below that
price.
rolledapairofdice:BirteEnglich,ThomasMussweiler,andFritzStrack,“PlayingDice
with Criminal Sentences: The Influence of Irrelevant Anchors on Experts’ Judicial
DecisionMaking,”PersonalityandSocialPsychologyBulletin32(2006):188–200.
NOLIMITPERPERSON:BrianWansink,RobertJ.Kent,andStephenJ.Hoch,“AnAnchoring
andAdjustmentModelofPurchaseQuantityDecisions,”JournalofMarketingResearch
35(1998):71–81.
resisttheanchoringeffect:AdamD.GalinskyandThomasMussweiler,“FirstOffersas
Anchors:TheRoleofPerspective-TakingandNegotiatorFocus,”JournalofPersonality
andSocialPsychology81(2001):657–69.
otherwise be much smaller: Greg Pogarsky and Linda Babcock, “Damage Caps,
MotivatedAnchoring,andBargainingImpasse,”JournalofLegalStudies30(2001):143–
59.
amount of damages: For an experimental demonstration, see Chris Guthrie, Jeffrey J.
Rachlinski,andAndrewJ.Wistrich,“JudgingbyHeuristic-CognitiveIllusionsinJudicial
DecisionMaking,”Judicature86(2002):44–50.
12:TheScienceofAvailability
“theeasewithwhich”:AmosTverskyandDanielKahneman,“Availability:AHeuristic
forJudgingFrequencyandProbability,”CognitivePsychology5(1973):207–32.
self-assessed contributions: Michael Ross and Fiore Sicoly, “Egocentric Biases in
Availability and Attribution,” Journal of Personality and Social Psychology 37 (1979):
322–36.
Amajoradvance:Schwarzetal.,“EaseofRetrievalasInformation.”
role of fluency: Sabine Stepper and Fritz Strack, “Proprioceptive Determinants of
EmotionalandNonemotionalFeelings,”JournalofPersonalityandSocialPsychology64
(1993):211–20.
experimentersdreamedup:Forareviewofthisareaofresearch,seeRainerGreifeneder,
HerbertBless,andMichelT.Pham,“WhenDoPeopleRelyonAffectiveandCognitive
FeelingsinJudgment?AReview,”PersonalityandSocialPsychologyReview15(2011):
107–41.
affect their cardiac health: Alexander Rotliman and Norbert Schwarz, “Constructing
Perceptions of Vulnerability: Personal Relevance and the Use of Experimental
InformationinHealthJudgments,”PersonalityandSocialPsychologyBulletin24(1998):
1053–64.
effortful task at the same time: Rainer Greifeneder and Herbert Bless, “Relying on
AccessibleContentVersusAccessibilityExperiences:TheCaseofProcessingCapacity,”
SocialCognition25(2007):853–81.
happyepisodeintheirlife:MarkusRuderandHerbertBless,“MoodandtheRelianceon
theEaseofRetrievalHeuristic,”JournalofPersonalityandSocialPsychology85(2003):
20–32.
low on a depression scale: Rainer Greifeneder and Herbert Bless, “Depression and
RelianceonEase-of-RetrievalExperiences,”European Journal of Social Psychology38
(2008):213–30.
knowledgeablenovices: Chezy Ofir et al., “Memory-Based Store Price Judgments: The
RoleofKnowledgeandShoppingExperience,”JournalofRetailing84(2008):414–23.
trueexperts:EugeneM.Caruso,“UseofExperiencedRetrievalEaseinSelfandSocial
Judgments,”JournalofExperimentalSocialPsychology44(2008):148–55.
faithinintuition:JohannesKellerandHerbertBless,“PredictingFutureAffectiveStates:
HowEaseofRetrievalandFaithinIntuitionModeratetheImpactofActivatedContent,”
EuropeanJournalofSocialPsychology38(2008):1–10.
if they are…powerful: Mario Weick and Ana Guinote, “When Subjective Experiences
Matter:PowerIncreasesRelianceontheEaseofRetrieval,”JournalofPersonalityand
SocialPsychology94(2008):956–70.
13:Availability,Emotion,andRisk
becauseofbraindamage:Damasio’sideaisknownasthe“somaticmarkerhypothesis”
andithasgatheredsubstantialsupport:AntonioR.Damasio,Descartes’Error:Emotion,
Reason, and the Human Brain (New York: Putnam, 1994). Antonio R. Damasio, “The
Somatic Marker Hypothesis and the Possible Functions of the Prefrontal Cortex,”
PhilosophicalTransactions:BiologicalSciences351(1996):141–20.
risksofeachtechnology:Finucaneetal.,“TheAffectHeuristicinJudgmentsofRisksand
Benefits.”PaulSlovic,MelissaFinucane,EllenPeters,andDonaldG.MacGregor,“The
Affect Heuristic,” in Thomas Gilovich, Dale Griffin, and Daniel Kahneman, eds.,
Heuristics and Biases (New York: Cambridge University Press, 2002), 397–420. Paul
Slovic,MelissaFinucane,EllenPeters,andDonaldG.MacGregor,“RiskasAnalysisand
Risk as Feelings: Some Thoughts About Affect, Reason, Risk, and Rationality,” Risk
Analysis 24 (2004): 1–12. Paul Slovic, “Trust, Emotion, Sex, Politics, and Science:
SurveyingtheRisk-AssessmentBattlefield,”RiskAnalysis19(1999):689–701.
British Toxicology Society: Slovic, “Trust, Emotion, Sex, Politics, and Science.” The
technologiesandsubstancesusedinthesestudiesarenotalternativesolutionstothesame
problem.Inrealisticproblems,wherecompetitivesolutionsareconsidered,thecorrelation
between costs and benefits must be negative; the solutions that have {ns problems,the
largestbenefitsarealsothemostcostly.Whetherlaypeopleandevenexpertsmightfailto
recognizethecorrectrelationshipeveninthosecasesisaninterestingquestion.
“wagstherationaldog”:JonathanHaidt,“TheEmotionalDogandItsRationalTail:A
Social Institutionist Approach to Moral Judgment,” Psychological Review 108 (2001):
814–34.
“‘Risk’ does not exist”: Paul Slovic, The Perception of Risk (Sterling, VA: EarthScan,
2000).
availabilitycascade:TimurKuranandCassR.Sunstein,“AvailabilityCascadesandRisk
Regulation,” Stanford Law Review 51 (1999): 683–768. CERCLA, the Comprehensive
EnvironmentalResponse,Compensation,andLiabilityAct,passedin1980.
nothinginbetween:PaulSlovic,whotestifiedfortheapplegrowersintheAlarcase,hasa
ratherdifferentview:“ThescarewastriggeredbytheCBS60Minutesbroadcastthatsaid
4,000childrenwilldieofcancer(noprobabilitiesthere)alongwithfrighteningpicturesof
bald children in a cancer ward—and many more incorrect statements. Also the story
exposed EPA’s lack of competence in attending to and evaluating the safety of Alar,
destroying trust in regulatory control. Given this, I think the public’s response was
rational.”(Personalcommunication,May11,2011.)
14:TomW’sSpecialty
“a shy poetry lover”: I borrowed this example from Max H. Bazerman and Don A.
Moore,JudgmentinManagerialDecisionMaking(NewYork:Wiley,2008).
always weighted more: Jonathan St. B. T. Evans, “Heuristic and Analytic Processes in
Reasoning,”BritishJournalofPsychology75(1984):451–68.
theoppositeeffect:NorbertSchwarzetal.,“BaseRates,Representativeness,andtheLogic
ofConversation:TheContextualRelevanceof‘Irrelevant’Information,”SocialCognition
9(1991):67–84.
toldtofrown:Alter,Oppenheimer,Epley,andEyre,“OvercomingIntuition.”
Bayes’srule: The simplest formof Bayes’s rule is inodds form, posterior odds =prior
odds×likelihoodratio,wheretheposterioroddsaretheodds(theratioofprobabilities)
fortwo competinghypotheses.Consider aproblem ofdiagnosis. Your friendhas tested
positiveforaseriousdisease.Thediseaseisrare:only1in600ofthecasessentinfor
testing actually has the disease. The test is fairly accurate. Its likelihood ratio is 25:1,
whichmeansthattheprobabilitythatapersonwhohasthediseasewilltestpositiveis25
timeshigherthantheprobabilityofafalsepositive.Testingpositiveisfrighteningnews,
buttheoddsthatyourfriendhasthediseasehaverisenonlyfrom1/600to25/600,andthe
probabilityis4%.
ForthehypothesisthatTomWisacomputerscientist,theprioroddsthatcorrespond
toabaserateof3%are(.03/.97=.031).Assumingalikelihoodratioof4(thedescription
is4timesaslikelyifTomWisacomputerscientistthanifheisnot),theposteriorodds
are 4 × . 031 = 12.4. From these odds you can { odes as l compute that the posterior
probabilityofTomWbeingacomputerscientistisnow11%(because12.4/112.4=.11).
15:Linda:LessisMore
theroleofheuristics:AmosTverskyandDanielKahneman,“ExtensionalVersusIntuitive
Reasoning: The Conjunction Fallacy in Probability Judgment,” Psychological Review
90(1983),293-315.
“a little homunculus”: Stephen Jay Gould, Bully for Brontosaurus (New York: Norton,
1991).
weakened or explained: See, among others, Ralph Hertwig and Gerd Gigerenzer, “The
‘ConjunctionFallacy’Revisited:HowIntelligentInferencesLookLikeReasoningErrors,”
JournalofBehavioralDecisionMaking12(1999):275–305;RalphHertwig,BjoernBenz,
andStefanKrauss,“TheConjunctionFallacyandtheManyMeaningsofAnd,”Cognition
108(2008):740–53.
settle our differences: Barbara Mellers, Ralph Hertwig, and Daniel Kahneman, “Do
Frequency Representations Eliminate Conjunction Effects? An Exercise in Adversarial
Collaboration,”PsychologicalScience12(2001):269–75.
16:CausesTrumpStatistics
correctansweris41%:ApplyingBayes’sruleinoddsform,theprioroddsaretheodds
fortheBluecabfromthebaserate,andthelikelihoodratioistheratiooftheprobability
ofthewitnesssayingthecabisBlueifitisBlue,dividedbytheprobabilityofthewitness
sayingthecabisBlueifitisGreen:posteriorodds=(.15/.85)×(.80/.20)=.706.Theodds
aretheratiooftheprobabilitythatthecabisBlue,dividedbytheprobabilitythatthecab
isGreen.ToobtaintheprobabilitythatthecabisBlue,wecompute:Probability(Blue)=
.706/1.706=.41.TheprobabilitythatthecabisBlueis41%.
nottoofarfromtheBayesian:AmosTverskyandDanielKahneman,“CausalSchemasin
Judgments Under Uncertainty,” in Progress in Social Psychology, ed. Morris Fishbein
(Hillsdale,NJ:Erlbaum,1980),49–72.
University of Michigan: Richard E. Nisbett and Eugene Borgida, “Attribution and the
PsychologyofPrediction,”JournalofPersonalityandSocialPsychology32(1975):932–
43.
relieved of responsibility: John M. Darley and Bibb Latane, “Bystander Intervention in
Emergencies:DiffusionofResponsibility,”JournalofPersonalityandSocialPsychology
8(1968):377–83.
17:RegressiontotheMean
help of the most brilliant statisticians: Michael Bulmer, Francis Galton: Pioneer of
HeredityandBiometry(Baltimore:JohnsHopkinsUniversityPress,2003).
standard scores: Researchers transform each original score into a standard score by
subtracting the mean and dividing the result by the standard deviation. Standard scores
have a mean of zero and a standard deviation of 1, can be compared across variables
(especially when the statistica {he deviatiol distributions of the original scores are
similar),andhavemanydesirablemathematicalproperties,whichGaltonhadtoworkout
tounderstandthenatureofcorrelationandregression.
correlationbetween parent and child:Thiswillnot be trueinanenvironmentin which
some children are malnourished. Differences in nutrition will become important, the
proportionofsharedfactorswilldiminish,andwithitthecorrelationbetweentheheight
ofparents andthe heightof children(unless the parents of malnourished childrenwere
alsostuntedbyhungerinchildhood).
heightandweight:Thecorrelationwascomputedforaverylargesampleofthepopulation
oftheUnitedStates(theGallup-HealthwaysWell-BeingIndex).
income and education: The correlation appears impressive, but I was surprised to learn
many years ago from the sociologist Christopher Jencks that if everyone had the same
education,theinequalityofincome(measuredbystandarddeviation)wouldbereduced
onlybyabout9%.Therelevantformulaisv(1–r2),whereristhecorrelation.
correlation and regression: This is true when both variables are measured in standard
scores—thatis,whereeachscoreistransformedbyremovingthemeananddividingthe
resultbythestandarddeviation.
confusing mere correlation with causation: Howard Wainer, “The Most Dangerous
Equation,”AmericanScientist95(2007):249–56.
18:TamingIntuitivePredictions
far more moderate: The proof of the standard regression as the optimal solution to the
prediction problem assumes that errors are weighted by the squared deviation from the
correctvalue.Thisistheleast-squarescriterion,whichiscommonlyaccepted.Otherloss
functionsleadtodifferentsolutions.
19:TheIllusionofUnderstanding
narrative fallacy: Nassim Nicholas Taleb, The Black Swan: The Impact of the Highly
Improbable(NewYork:RandomHouse,2007).
oneattributethatisparticularlysignificant:.
throwingtheball:MichaelLewis,Moneyball:TheArtofWinninganUnfairGame(New
York:Norton,2003).
sell their company:SethWeintraub, “Excite PassedUp BuyingGoogle for$750,000 in
1999,”Fortune,September29,2011.
everfeltdifferently:RichardE.NisbettandTimothyD.Wilson,“TellingMoreThanWe
CanKnow:VerbalReportsonMentalProcesses,”PsychologicalReview84(1977):231–
59.
UnitedStatesandtheSovietUnion:BaruchFischhoffandRuthBeyth,“IKnewItWould
Happen:RememberedProbabilitiesofOnceFutureThings,”OrganizationalBehaviorand
HumanPerformance13(1975):1–16.
qualityofadecision:JonathanBaronandJohnC.Hershey,“OutcomeBiasinDecision{s
iiv>Evaluation,”JournalofPersonalityandSocialPsychology54(1988):569–79.
shouldhavehiredthemonitor:KimA.KaminandJeffreyRachlinski,“ExPost?ExAnte:
Determining Liability in Hindsight,” Law and Human Behavior 19 (1995): 89–104.
Jeffrey J. Rachlinski, “A Positive Psychological Theory of Judging in Hindsight,”
UniversityofChicagoLawReview65(1998):571–625.
tidbitofintelligence:JeffreyGoldberg,“LetterfromWashington:Woodwardvs.Tenet,”
NewYorker,May21,2007,35–38.AlsoTimWeiner,LegacyofAshes:TheHistoryofthe
CIA (New York: Doubleday, 2007); “Espionage: Inventing the Dots,” Economist,
November3,2007,100.
reluctancetotakerisks:PhilipE.Tetlock,“Accountability:TheNeglectedSocialContext
ofJudgmentandChoice,”ResearchinOrganizationalBehavior7(1985):297–332.
beforetheircurrentappointment:MarianneBertrandandAntoinetteSchoar,“Managing
withStyle:The EffectofManagers on FirmPolicies,”Quarterly Journal of Economics
118(2003):1169–1208.NickBloomandJohnVanReenen,“MeasuringandExplaining
ManagementPracticesAcrossFirmsandCountries,”QuarterlyJournalofEconomics122
(2007):1351–1408.
“Howoftenwillyoufind…”:IamindebtedtoProfessorJamesH.SteigerofVanderbilt
University, who developed an algorithm that answers this question, under plausible
assumptions. Steiger’s analysis shows that correlations of .20 and .40 are associated,
respectively,withinversionratesof43%and37%.
hispenetratingbook:TheHaloEffectwaspraisedasoneofthebestbusinessbooksofthe
yearbyboththeFinancialTimesandTheWallStreetJournal:PhilRosenzweig,TheHalo
Effect:…and the Eight Other Business Delusions That Deceive Managers (New York:
Simon&Schuster,2007).SeealsoPaulOlkandPhilRosenzweig,“TheHaloEffectand
theChallengeofManagementInquiry:ADialogBetweenPhilRosenzweigandPaulOlk,”
JournalofManagementInquiry19(2010):48–54.
“a visionary company”: James C. Collins and Jerry I. Porras, Built to Last: Successful
HabitsofVisionaryCompanies(NewYork:Harper,2002).
flipofacoin:In fact,even if you were theCEO yourself,your forecastswould not be
impressivelyreliable;theextensiveresearchoninsidertradingshowsthatexecutivesdo
beatthemarketwhentheytradetheirownstock,butthemarginoftheiroutperformanceis
barely enough to cover the costs of trading. See H. Nejat Seyhun, “The Information
Content of Aggregate Insider Trading,” Journal of Business 61 (1988): 1–24; Josef
Lakonishok and Inmoo Lee, “Are Insider Trades Informative?” Review of Financial
Studies14(2001):79–111;ZahidIqbalandShekarShetty,“AnInvestigationofCausality
Between Insider Transactions and Stock Returns,” Quarterly Review of Economics and
Finance42(2002):41–57.
InSearchofExcellence:Rosenz{lenlatweig,TheHaloEffect.
“Most Admired Companies”: Deniz Anginer, Kenneth L. Fisher, and Meir Statman,
“StocksofAdmiredCompaniesandDespisedOnes,”workingpaper,2007.
regressiontothemean:JasonZweigobservesthatthelackofappreciationforregression
hasdetrimentalimplicationsfortherecruitmentofCEOs.Strugglingfirmstendtoturnto
outsiders,recruitingCEOsfromcompanieswithhighrecentreturns.TheincomingCEO
thengetscredit,atleasttemporarily,forhisnewfirm’ssubsequentimprovement.(Mean-
while, his replacement at his former firm is now struggling, leading the new bosses to
believethattheydefinitelyhired“therightguy.”)AnytimeaCEOjumpsship,thenew
companymustbuyouthisstake(instockandoptions)athisoldfirm,settingabaseline
forfuturecompensationthathasnothingtodowithperformanceatthenewfirm.Tensof
millions of dollars in compensation get awarded for “personal” achievements that are
driven mainly by regression and halo effects (personal communication, December 29,
2009).
20:TheIllusionofValidity
thisstartlingconclusion:BradM.BarberandTerranceOdean,“TradingIsHazardousto
Your Wealth: The Common Stock Investment Performance of Individual Investors,”
JournalofFinance55(2002):773–806.
men acted on their useless ideas: Brad M. Barber and Terrance Odean, “Boys Will Be
Boys: Gender, Overconfidence, and Common Stock Investment,” Quarterly Journal of
Economics116(2006):261–92.
selling“winners”:This“dispositioneffect”isdiscussedfurther.
respondingtonews:BradM.BarberandTerranceOdean,“AllThatGlitters:TheEffectof
Attention and News on the Buying Behavior of Individual and Institutional Investors,”
ReviewofFinancialStudies21(2008):785–818.
wealthfromamateurs:ResearchonstocktradesinTaiwanconcludedthatthetransferof
wealth from individuals to financial institutions amounts to a staggering 2.2% of GDP:
BradM.Barber,Yi-TsungLee,Yu-JaneLiu,andTerranceOdean,“JustHowMuchDo
IndividualInvestorsLosebyTrading?”ReviewofFinancialStudies22(2009):609–32.
underperformtheoverallmarket:JohnC.Bogle,CommonSenseonMutualFunds:New
ImperativesfortheIntelligentInvestor(NewYork:Wiley,2000),213.
persistentdifferences in skill: Mark Grinblatt and Sheridan Titman, “The Persistence of
MutualFundPerformance,”JournalofFinance42(1992):1977–84.EdwinJ.Eltonetal.,
“The Persistence of Risk-Adjusted Mutual Fund Performance,” Journal of Business 52
(1997): 1–33. Edwin Elton et al., “Efficiency With Costly Information: A Re-
interpretation of Evidence from Managed Portfolios,” Review of Financial Studies 6
(1993):1–21.
“In this age of academic hyperspecialization”: Philip E. Tetlock, Expert Political
Judgment:>HowGoodisIt?HowCanWeKnow?(Princeton:PrincetonUniversityPress,
2005),233.
21:Intuitionsvs.Formulas
“There is no controversy”: Paul Meehl, “Causes and Effects of My Disturbing Little
Book,”JournalofPersonalityAssessment50(1986):370–75.
afactorof10ormore:Duringthe1990–1991auctionseason,forexample,thepricein
Londonofacaseof1960ChateauLatouraveraged$464;acaseofthe1961vintage(one
ofthebestever)fetchedanaverageof$5,432.
Experienced radiologists: Paul J. Hoffman, Paul Slovic, and Leonard G. Rorer, “An
Analysis-of-VarianceModelfortheAssessmentofConfiguralCueUtilizationinClinical
Judgment,”PsychologicalBulletin69(1968):338–39.
internal corporate audits: Paul R. Brown, “Independent Auditor Judgment in the
EvaluationofInternalAuditFunctions,”JournalofAccountingResearch21(1983):444–
55.
41 separate studies: James Shanteau, “Psychological Characteristics and Strategies of
ExpertDecisionMakers,”ActaPsychologica68(1988):203–15.
successive food breaks: Danziger, Levav, and Avnaim-Pesso, “Extraneous Factors in
JudicialDecisions.”
lowering validity: Richard A. DeVaul et al., “Medical-School Performance of Initially
RejectedStudents,”JAMA257(1987):47–51.JasonDanaandRobynM.Dawes,“Belief
intheUnstructuredInterview:ThePersistenceofanIllusion,”workingpaper,Department
of Psychology, University of Pennsylvania, 2011. William M. Grove et al., “Clinical
Versus Mechanical Prediction: A Meta-Analysis,” Psychological Assessment 12 (2000):
19–30.
Dawes’s famous article: Robyn M. Dawes, “The Robust Beauty of Improper Linear
ModelsinDecisionMaking,”AmericanPsychologist34(1979):571–82.
notaffectedbyaccidentsofsampling:JasonDanaandRobynM.Dawes,“TheSuperiority
of Simple Alternatives to Regression for Social Science Predictions,” Journal of
EducationalandBehavioralStatistics29(2004):317–31.
Dr.Apgar:VirginiaApgar,“AProposalforaNewMethodofEvaluationoftheNewborn
Infant,”CurrentResearchesinAnesthesiaandAnalgesia32(1953):260–67.Mieczyslaw
Finster and Margaret Wood, “The Apgar Score Has Survived the Test of Time,”
Anesthesiology102(2005):855–57.
virtuesofchecklists:AtulGawande,The Checklist Manifesto: How toGetThingsRight
(NewYork:MetropolitanBooks,2009).
organic fruit: Paul Rozin, “The Meaning of ‘Natural’: Process More Important than
Content,”PsychologicalScience16(2005):652–58.
2{ce
moderated by an arbiter: Mellers, Hertwig, and Kahneman, “Do Frequency
RepresentationsEliminateConjunctionEffects?”
articulatedthisposition:Klein,SourcesofPower.
kouros:TheGettyMuseuminLosAngelesbringsintheworld’sleadingexpertsonGreek
sculpturetoviewakouros—amarblestatueofastridingboy—thatitisabouttobuy.One
after another, the experts react with what one calls “intuitive repulsion”—a powerful
hunchthatthekourosisnot2,500yearsoldbutamodernfake.Noneoftheexpertscan
immediatelysaywhytheythinkthesculptureisaforgery.Theclosestanyofthemcould
come to a rationale is an Italian art historian’s complaint that something—he does not
knowexactly what—“seemedwrong”with thestatue’s fingernails.Afamous American
expertsaidthatthefirstthoughtthatcametohismindwasthewordfresh,andaGreek
expert flatly stated, “Anyone who has ever seen a sculpture coming out of the ground
could tell that that thing has never been in the ground.” The lack of agreement on the
reasonsforthesharedconclusionisstriking,andrathersuspect.
admired as a hero: Simon was one of the towering intellectual figures of the twentieth
century.Hewroteaclassicondecisionmakinginorganizationswhilestillinhistwenties,
andamongmanyotherachievementshewentontobeoneofthefoundersofthefieldof
artificialintelligence,aleaderincognitivescience,aninfluentialstudentoftheprocessof
scientific discovery, a forerunner of behavioral economics and, almost incidentally, a
Nobellaureateineconomics.
“nothinglessthanrecognition”:Simon,“WhatIsanExplanationofBehavior?”DavidG.
Myers,Intuition:ItsPowersandPerils(NewHaven:YaleUniversityPress,2002),56.
“withoutknowinghowheknows”:SeymourEpstein,“DemystifyingIntuition:WhatItIs,
WhatItDoes,HowItDoesIt,”PsychologicalInquiry21(2010):295–312.
10,000hours:Foer,MoonwalkingwithEinstein.
23:TheOutsideView
insideviewandtheoutsideview:Thelabelsareoftenmisunderstood.Numerousauthors
believed that the correct terms were “insider view” and “outsider view,” which are not
evenclosetowhatwehadinmind.
very different answers: Dan Lovallo and Daniel Kahneman, “Timid Choices and Bold
Forecasts:ACognitivePerspectiveonRiskTaking,”ManagementScience39(1993):17–
31. Daniel Kahneman and Dan Lovallo, “Delusions of Success: How Optimism
UnderminesExecutives’Decisions,”HarvardBusinessReview81(2003):56–63.
“Pallid”statisticalinformation:RichardE.NisbettandLeeD.Ross,HumanInference:
Strategies and Shortcomings of Social Judgment (Englewood Cliffs, NJ: Prentice-Hall,
1980).
impersonality of procedures: Fo {i>How Doctors Think (New York: Mariner Books,
2008),6.
planningfallacy:DanielKahnemanandAmosTversky,“IntuitivePrediction:Biasesand
CorrectiveProcedures,”ManagementScience12(1979):313–27.
Scottish Parliament building: Rt. Hon. The Lord Fraser of Carmyllie, “The Holyrood
Inquiry, Final Report,” September 8, 2004,
www.holyroodinquiry.org/FINAL_report/report.htm.
didnotbecomemorereliantonit:BrentFlyvbjerg,MetteK.SkamrisHolm,andSørenL.
Buhl,“How(In)accurateAreDemandForecastsinPublicWorksProjects?”Journalofthe
AmericanPlanningAssociation71(2005):131–46.
surveyofAmericanhomeowners:“2002Costvs.ValueReport,”Remodeling,November
20,2002.
completiontimes: Brent Flyvbjerg, “From Nobel Prize to Project Management: Getting
RisksRight,”ProjectManagementJournal37(2006):5–15.
sunk-costfallacy:HalR.ArkesandCatherineBlumer,“ThePsychologyofSunkCost,”
Organizational Behavior and Human Decision Processes 35 (1985): 124–40. Hal R.
ArkesandPeterAyton,“TheSunkCostandConcordeEffects:AreHumansLessRational
ThanLowerAnimals?”PsychologicalBulletin125(1998):591–600.
24:TheEngineofCapitalism
you already feel fortunate: Miriam A. Mosing et al., “Genetic and Environmental
InfluencesonOptimismandItsRelationshiptoMentalandSelf-RatedHealth:AStudyof
Aging Twins,” Behavior Genetics 39 (2009): 597–604. David Snowdon, Aging with
Grace: Whatthe Nun Study Teaches Us About Leading Longer, Healthier, and More
MeaningfulLives(NewYork:BantamBooks,2001).
brightsideofeverything:ElaineFox,AnnaRidgewell,andChrisAshwin,“Lookingon
the Bright Side: Biased Attention and the Human Serotonin Transporter Gene,”
ProceedingsoftheRoyalSocietyB276(2009):1747–51.
“triumphofhopeoverexperience”:ManjuPuriandDavidT.Robinson,“Optimismand
EconomicChoice,”JournalofFinancialEconomics86(2007):71–99.
more sanguine than midlevel managers: Lowell W. Busenitz and Jay B. Barney,
“Differences Between Entrepreneurs and Managers in Large Organizations: Biases and
HeuristicsinStrategicDecision-Making,”JournalofBusinessVenturing12(1997):9–30.
admirationofothers:Entrepreneurswhohavefailedaresustainedintheirconfidenceby
the probably mistaken belief that they have learned a great deal from the experience.
GavinCassarand Justin Craig,“AnInvestigationof Hindsight BiasinNascentVenture
Activity,”JournalofBusinessVenturing24({>
influenceonthelivesofothers:KeithM.HmieleskiandRobertA.Baron,“Entrepreneurs’
OptimismandNewVenturePerformance:ASocialCognitivePerspective,”Academyof
Management Journal 52 (2009): 473–88. Matthew L. A. Hayward, Dean A. Shepherd,
and Dale Griffin, “A Hubris Theory of Entrepreneurship,” Management Science 52
(2006):160–72.
chance of failing was zero: Arnold C. Cooper, Carolyn Y. Woo, and William C.
Dunkelberg, “Entrepreneurs’ Perceived Chances for Success,” Journal of Business
Venturing3(1988):97–108.
given the lowest grade: Thomas Astebro and Samir Elhedhli, “The Effectiveness of
SimpleDecisionHeuristics:ForecastingCommercialSuccessforEarly-StageVentures,”
ManagementScience52(2006):395–409.
widespread, stubborn, and costly: Thomas Astebro, “The Return to Independent
Invention: Evidence of Unrealistic Optimism, Risk Seeking or Skewness Loving?”
EconomicJournal113(2003):226–39.
bet small amounts of money: Eleanor F. Williams and Thomas Gilovich, “Do People
ReallyBelieveTheyAreAboveAverage?”JournalofExperimentalSocialPsychology44
(2008):1121–28.
“hubris hypothesis”: Richard Roll, “The Hubris Hypothesis of Corporate Takeovers,”
JournalofBusiness59(1986):197–216,part1.Thisremarkableearlyarticlepresenteda
behavioral analysis of mergers and acquisitions that abandoned the assumption of
rationality,longbeforesuchanalysesbecamepopular.
“value-destroying mergers”: Ulrike Malmendier and Geoffrey Tate, “Who Makes
Acquisitions? CEO Overconfidence and the Market’s Reaction,” Journal of Financial
Economics89(2008):20–43.
“engage in earnings management”: Ulrike Malmendier and Geoffrey Tate, “Superstar
CEOs,”QuarterlyJournalofEconomics24(2009),1593–1638.
self-aggrandizement to a cognitive bias: Paul D. Windschitl, Jason P. Rose, Michael T.
Stalk-fleet, and Andrew R. Smith, “Are People Excessive or Judicious in Their
Egocentrism? A Modeling Approach to Understanding Bias and Accuracy in People’s
Optimism,”JournalofPersonalityandSocialPsychology95(2008):252–73.
averageoutcomeisaloss:Aformofcompetitionneglecthasalsobeenobservedinthe
timeofdayatwhichsellersoneBaychoosetoendtheirauctions.Theeasyquestionis:At
whattimeisthetotalnumberofbiddersthehighest?Answer:around7:00p.m.EST.The
question sellers should answer is harder: Considering how many other sellers end their
auctions during peak hours, at what time will there be the most bidders looking at my
auction? The answer: around noon, when the number of bidders is large relative to the
numberofsellers.The sellers who rememberthecompetitionandavoidprimetimeget
higher prices. Uri Simonsohn, “eBay’s Crowded Evenings: Competition Neglect in
MarketEntryDecisions,”ManagementScience56(2010):1060–73.
“diagnosisantemortem”:EtaS.BernerandMarkL.Graber,“OverconfidenceasaCause
ofDiagnosticErrorinMedicine,”AmericanJournalofMedicine121(2008):S2–S23.
“disclosinguncertaintytopatients”:PatCroskerryandGeoffNorman,“Overconfidence
inClinicalDecisionMaking,”AmericanJournalofMedicine121(2008):S24–S29.
backgroundofrisktaking:KahnemanandLovallo,“TimidChoicesandBoldForecasts.”
Royal Dutch Shell: J. Edward Russo and Paul J. H. Schoemaker, “Managing
Overconfidence,”SloanManagementReview33(1992):7–17.
25:Bernoulli’sErrors
Mathematical Psychology: Clyde H. Coombs, Robyn M. Dawes, and Amos Tversky,
MathematicalPsychology:AnElementaryIntroduction(EnglewoodCliffs,NJ:Prentice-
Hall,1970).
for the rich and for the poor: This rule applies approximately to many dimensions of
sensationandperception.ItisknownasWeber’slaw,aftertheGermanphysiologistErnst
Heinrich Weber, who discovered it. Fechner drew on Weber’s law to derive the
logarithmicpsychophysicalfunction.
$10millionfrom$100million:Bernoulli’sintuitionwascorrect,andeconomistsstilluse
thelogofincomeorwealthinmanycontexts.Forexample,whenAngusDeatonplotted
the average life satisfaction of residents of many countries against the GDP of these
countries,heusedthelogarithmofGDPasameasureofincome.Therelationship,itturns
out,isextremelyclose:Residentsofhigh-GDPcountriesaremuchmoresatisfiedwiththe
qualityoftheirlivesthanareresidentsofpoorcountries,andadoublingofincomeyields
approximatelythesameincrementofsatisfactioninrichandpoorcountriesalike.
“St. Petersburg paradox”: Nicholas Bernoulli, a cousin of Daniel Bernoulli, asked a
questionthatcanbeparaphrasedasfollows:“Youareinvitedtoagameinwhichyoutoss
a coin repeatedly. You receive $2 if it shows heads, and the prize doubles with every
successive toss that shows heads. The game ends when the coin first shows tails. How
much would you pay for an opportunity to play that game?” People do not think the
gambleisworthmorethanafewdollars,althoughitsexpectedvalueisinfinite—because
theprizekeepsgrowing,theexpectedvalueis$1foreachtoss,toinfinity.However,the
utility of the prizes grows much more slowly, which explains why the gamble is not
attractive.
“historyofone’swealth”:OtherfactorscontributedtothelongevityofBernoulli’stheory.
Oneisthatitisnaturaltoformulatechoicesbetweengamblesintermsofgains,ormixed
gains and losses. Not many people thought about choices in which all options are bad,
althoughwewerebynomeansthefirsttoobserveriskseeking.Anotherfactthatfavors
Bernoulli’stheoryisthatthinkingintermsoffinalstatesofwealthandignoringthepastis
oftenaveryreasonablethingtodo.Economistsweretraditionallyconcernedwithrational
choices,andBernoulli’smodelsuitedtheirgoal.
26:ProspectTheory
ast=“2%”>
subjectivevalueofwealth:StanleyS.Stevens,“ToHonorFechnerandRepealHisLaw,”
Science133(1961):80–86.Stevens,Psychophysics.
The three principles: Writing this sentence reminded me that the graph of the value
functionhasalreadybeenusedasanemblem.EveryNobellaureatereceivesanindividual
certificate with a personalized drawing, which is presumably chosen by the committee.
Myillustrationwasastylizedrenditionoffigure10.
“lossaversionratio”:Thelossaversionratioisoftenfoundtobeintherangeof1.5and
2.5: Nathan Novemsky and Daniel Kahneman, “The Boundaries of Loss Aversion,”
JournalofMarketingResearch42(2005):119–28.
emotional reaction to losses: Peter Sokol-Hessner et al., “Thinking Like a Trader
SelectivelyReducesIndividuals’LossAversion,”PNAS106(2009):5035–40.
Rabin’stheorem:Forseveralconsecutiveyears,Igaveaguestlectureintheintroductory
financeclassofmycolleagueBurtonMalkiel.IdiscussedtheimplausibilityofBernoulli’s
theory each year. I noticed a distinct change in my colleague’s attitude when I first
mentioned Rabin’s proof. He was now prepared to take the conclusion much more
seriouslythaninthepast.Mathematicalargumentshaveadefinitivequalitythatismore
compelling than appeals to common sense. Economists are particularly sensitive to this
advantage.
rejectsthatgamble:Theintuitionoftheproofcanbeillustratedbyanexample.Suppose
anindividual’swealthisW,andsherejectsagamblewithequalprobabilitiestowin$11
or lose $10. If the utility function for wealth is concave (bent down), the preference
impliesthatthevalueof$1hasdecreasedbyover9%overanintervalof$21!Thisisan
extraordinarilysteepdeclineandtheeffectincreasessteadilyasthegamblesbecomemore
extreme.
“Evenalousylawyer”:MatthewRabin,“RiskAversionandExpected-UtilityTheory:A
CalibrationTheorem,”Econometrica68(2000):1281–92.MatthewRabinandRichardH.
Thaler,“Anomalies:RiskAversion,”JournalofEconomicPerspectives15(2001):219–
32.
economistsandpsychologists:Severaltheoristshaveproposedversionsofregrettheories
thatarebuiltontheideathatpeopleareabletoanticipatehowtheirfutureexperiences
willbeaffectedbytheoptionsthatdidnotmaterializeand/orbythechoicestheydidnot
make: David E. Bell, “Regret in Decision Making Under Uncertainty,” Operations
Research 30(1982): 961–81. GrahamLoomes and RobertSugden, “Regret Theory:An
AlternativetoRationalChoiceUnderUncertainty,”EconomicJournal92(1982):805–25.
BarbaraA.Mellers,“ChoiceandtheRelativePleasureofConsequences,”Psychological
Bulletin 126 (2000): 910–24. Barbara A. Mellers, Alan Schwartz, and Ilana Ritov,
“Emotion-Based Choice,” Journal of Experimental Psychology—General 128 (1999):
332–45. Decision makers’ choices between gambles depend on whether they expect to
knowtheoutcomeofthegambletheydidnotchoose.IlanaRitov,“ProbabilityofRegret:
Anticipation of Uncertainty Resolution in Choice,” Organiz {an>y did not ational
BehaviorandHumanDecisionProcesses66(1966):228–36.
27:TheEndowmentEffect
Whatismissingfromthefigure:Atheoreticalanalysisthatassumeslossaversionpredicts
a pronounced kink of the indifference curve at the reference point: Amos Tversky and
DanielKahneman,“LossAversioninRisklessChoice:AReference-DependentModel,”
QuarterlyJournalofEconomics106(1991):1039–61.JackKnetschobservedthesekinks
in an experimental study: “Preferences and Nonreversibility of Indifference Curves,”
JournalofEconomicBehavior&Organization17(1992):131–39.
periodofoneyear:AlanB.KruegerandAndreasMueller,“JobSearchandJobFindingin
a Period of Mass Unemployment: Evidence from High-Frequency Longitudinal Data,”
workingpaper,PrincetonUniversityIndustrialRelationsSection,January2011.
didnotownthebottle:Technically,thetheoryallowsthebuyingpricetobeslightlylower
thanthesellingpricebecauseofwhateconomistscallan“incomeeffect”:Thebuyerand
the seller are not equally wealthy, because the seller has an extra bottle. However, the
effectinthiscaseisnegligiblesince$50isaminutefractionoftheprofessor’s wealth.
Thetheorywouldpredictthatthisincomeeffectwouldnotchangehiswillingnesstopay
byevenapenny.
wouldbepuzzledbyit:TheeconomistAlanKruegerreportedonastudyheconductedon
theoccasionoftakinghisfathertotheSuperBowl:“Weaskedfanswhohadwontheright
tobuyapairofticketsfor$325or$400eachinalotterywhethertheywouldhavebeen
willingtopay$3,000aticketiftheyhadlostinthelotteryandwhethertheywouldhave
sold their tickets if someone had offered them $3,000 apiece. Ninety-four percent said
theywouldnothaveboughtfor$3,000,andninety-twopercentsaidtheywouldnothave
soldatthatprice.”Heconcludesthat“rationalitywasinshortsupplyattheSuperBowl.”
AlanB.Krueger,“SupplyandDemand:AnEconomistGoestotheSuperBowl,”Milken
InstituteReview:AJournalofEconomicPolicy3(2001):22–29.
givingup abottleofnicewine:Strictlyspeaking, loss aversionreferstotheanticipated
pleasureandpain,whichdeterminechoices.Theseanticipationscouldbewronginsome
cases. Deborah A. Kermer et al., “Loss Aversion Is an Affective Forecasting Error,”
PsychologicalScience17(2006):649–53.
markettransactions:NovemskyandKahneman,“TheBoundariesofLossAversion.”
halfofthetokenswillchangehands:Imaginethatalltheparticipantsareorderedinaline
by the redemption value assigned to them. Now randomly allocate tokens to half the
individualsintheline.Halfofthepeopleinthefrontofthelinewillnothaveatoken,and
halfofthepeopleattheendofthelinewillownone.Thesepeople(halfofthetotal)are
expectedtomovebytradingplaceswitheachother,sothatintheendeveryoneinthefirst
halfofthelinehasatoken,andnoonebehindthemdoes.
Brainrecordings:BrianKnutsonetal.,“NeuralAntecedentsoftheEndowmentEffect,”
Neuron 58 (2008): 814–22. Brian Knutson an {an utson et ad Stephanie M. Greer,
“Anticipatory Affect: Neural Correlates and Consequences for Choice,” Philosophical
TransactionsoftheRoyalSocietyB363(2008):3771–86.
riskless and risky decisions: A review of the price of risk, based on “international data
from16differentcountriesduringover100years,”yieldedanestimateof2.3,“instriking
agreement with estimates obtained in the very different methodology of laboratory
experimentsofindividualdecision-making”:MosheLevy,“LossAversionandthePriceof
Risk,”QuantitativeFinance10(2010):1009–22.
effect of price increases: Miles O. Bidwel, Bruce X. Wang, and J. Douglas Zona, “An
Analysis of Asymmetric Demand Response to Price Changes: The Case of Local
Telephone Calls,” Journal of Regulatory Economics 8 (1995): 285–98. Bruce G. S.
Hardie, Eric J. Johnson, and Peter S. Fader, “Modeling Loss Aversion and Reference
DependenceEffectsonBrandChoice,”MarketingScience12(1993):378–94.
illustrate the power of these concepts: Colin Camerer, “Three Cheers—Psychological,
Theoretical, Empirical—for Loss Aversion,” Journal of Marketing Research 42 (2005):
129–33. Colin F. Camerer, “Prospect Theory in the Wild: Evidence from the Field,” in
Choices, Values, and Frames, ed. Daniel Kahneman and Amos Tversky (New York:
RussellSageFoundation,2000),288–300.
condoapartmentsinBoston:DavidGenesoveandChristopherMayer,“LossAversionand
Seller Behavior: Evidence from the Housing Market,” Quarterly Journal of Economics
116(2001):1233–60.
effect of trading experience: John A. List, “Does Market Experience Eliminate Market
Anomalies?”QuarterlyJournalofEconomics118(2003):47–71.
Jack Knetsch also: Jack L. Knetsch, “The Endowment Effect and Evidence of
NonreversibleIndifferenceCurves,”AmericanEconomicReview79(1989):1277–84.
ongoing debate about the endowment effect: Charles R. Plott and Kathryn Zeiler, “The
Willingness to Pay–Willingness to Accept Gap, the ‘Endowment Effect,’ Subject
Misconceptions, and Experimental Procedures for Eliciting Valuations,” American
EconomicReview95(2005):530–45.CharlesPlott,aleadingexperimentaleconomist,has
been very skeptical of the endowment effect and has attempted to show that it is not a
“fundamental aspect of human preference” but rather an outcome of inferior technique.
PlottandZeilerbelievethatparticipantswhoshowtheendowmenteffectareundersome
misconceptionaboutwhattheirtruevaluesare,andtheymodifiedtheproceduresofthe
originalexperimentstoeliminatethemisconceptions.Theydevisedanelaboratetraining
procedureinwhichtheparticipantsexperiencedtherolesofbothbuyersandsellers,and
were explicitly taught to assess their true values. As expected, the endowment effect
disappeared. Plott and Zeiler view their method as an important improvement of
technique. Psychologists would consider the method severely deficient, because it
communicates to the participants a message of what the experimenters consider
appropriatebehavior,whichhappenstocoincidewiththeexperimenters’theory.Plottand
Zeiler’sfavoredversionofKne{ers):tsch’sexchangeexperimentissimilarlybiased:It
doesnotallowtheownerofthegoodtohavephysicalpossessionofit,whichiscrucialto
theeffect.SeeCharlesR.PlottandKathrynZeiler,“ExchangeAsymmetriesIncorrectly
Interpreted as Evidence of Endowment Effect Theory and Prospect Theory?” American
EconomicReview97(2007):1449–66.Theremaybeanimpassehere,whereeachside
rejectsthemethodsrequiredbytheother.
People who are poor: In their studies of decision making under poverty, Eldar Shafir,
Sendhil Mullainathan, and their colleagues have observed other instances in which
poverty induces economic behavior that is in some respects more realistic and more
rationalthanthatofpeoplewhoarebetteroff.Thepooraremorelikelytorespondtoreal
outcomesthanto theirdescription.MarianneBertrand,SendhilMullainathan,andEldar
Shafir, “Behavioral Economics and Marketing in Aid of Decision Making Among the
Poor,”JournalofPublicPolicy&Marketing25(2006):8–23.
intheUnitedStatesandintheUK:Theconclusionthatmoneyspentonpurchasesisnot
experiencedasalossismorelikelytobetrueforpeoplewhoarerelativelywell-off.The
keymaybewhetheryouareawarewhenyoubuyonegoodthatyouwillnotbeunableto
affordanothergood.NovemskyandKahneman,“TheBoundariesofLossAversion.”Ian
Bateman et al., “Testing Competing Models of Loss Aversion: An Adversarial
Collaboration,”JournalofPublicEconomics89(2005):1561–80.
28:BadEvents
heartbeataccelerated:PaulJ.Whalenetal.,“HumanAmygdalaResponsivitytoMasked
Fearful Eye Whites,” Science 306 (2004): 2061. Individuals with focal lesions of the
amygdalashowedlittleornolossaversionintheirriskychoices:BenedettoDeMartino,
Colin F. Camerer, and Ralph Adolphs, “Amygdala Damage Eliminates Monetary Loss
Aversion,”PNAS107(2010):3788–92.
bypassing the visual cortex: Joseph LeDoux, The Emotional Brain: The Mysterious
UnderpinningsofEmotionalLife(NewYork:Touchstone,1996).
processed faster: Elaine Fox et al., “Facial Expressions of Emotion: Are Angry Faces
DetectedMoreEfficiently?”Cognition&Emotion14(2000):61–92.
“popsout”:ChristineHansenandRanaldHansen,“FindingtheFaceintheCrowd:An
AngerSuperiorityEffect,”JournalofPersonalityandSocialPsychology54(1988):917–
24.
“acceptable/unacceptable”:JosJ.A.VanBerkumetal.,“RightorWrong?TheBrain’s
Fast Response to Morally Objectionable Statements,” Psychological Science 20 (2009):
1092–99.
negativitydominance:PaulRozinandEdwardB.Royzman,“NegativityBias,Negativity
Dominance,andContagion,”Personality andSocialPsychology Review5(2001):296–
320.
resistanttodisconfirmation:RoyF.Baumeister,EllenBratslavsky,CatrinFinkenauer,and
KathleenD.Vohs,“BadIsStrongerThanGood,”ReviewofGeneralPsychology5(200
{/spFac1):323.
biologically significant improvement: Michel Cabanac, “Pleasure: The Common
Currency,”JournalofTheoreticalBiology155(1992):173–200.
not equally powerful: Chip Heath, Richard P. Larrick, and George Wu, “Goals as
ReferencePoints,”CognitivePsychology38(1999):79–109.
rain-drenched customers: Colin Camerer, Linda Babcock, George Loewenstein, and
Richard Thaler, “Labor Supply of New York City Cabdrivers: One Day at a Time,”
Quarterly Journal of Economics 112 (1997): 407–41. The conclusions of this research
havebeenquestioned:HenryS.Farber,“IsTomorrowAnotherDay?TheLaborSupplyof
NewYorkCabDrivers,”NBERWorkingPaper9706,2003.Aseriesofstudiesofbicycle
messengersinZurichprovidesstrongevidencefortheeffectofgoals,inaccordwiththe
originalstudyofcabdrivers:ErnstFehrandLorenzGoette,“DoWorkersWorkMoreif
WagesAreHigh?EvidencefromaRandomizedFieldExperiment,”AmericanEconomic
Review97(2007):298–317.
communicateareference point:DanielKahneman, “ReferencePoints, Anchors, Norms,
andMixedFeelings,”OrganizationalBehaviorandHumanDecisionProcesses51(1992):
296–312.
“wins the contest”: John Alcock, Animal Behavior: An Evolutionary Approach
(Sunderland, MA: Sinauer Associates, 2009), 278–84, cited by Eyal Zamir, “Law and
Psychology: The Crucial Role of Reference Points and Loss Aversion,” working paper,
HebrewUniversity,2011.
merchants,employers,andlandlords:DanielKahneman,JackL.Knetsch,andRichardH.
Thaler, “Fairness as a Constraint on Profit Seeking: Entitlements in the Market,” The
AmericanEconomicReview76(1986):728–41.
fairnessconcernsareeconomicallysignificant:ErnstFehr,LorenzGoette,andChristian
Zehnder,“ABehavioralAccountoftheLaborMarket:TheRoleofFairnessConcerns,”
AnnualReviewofEconomics1(2009):355–84.EricT.AndersonandDuncanI.Simester,
“Price Stickiness and Customer Antagonism,” Quarterly Journal of Economics 125
(2010):729–65.
altruisticpunishmentisaccompanied:DominiquedeQuervainetal.,“TheNeuralBasisof
AltruisticPunishment,”Science305(2004):1254–58.
actuallossesandforegonegains:DavidCohenandJackL.Knetsch,“JudicialChoiceand
DisparitiesBetweenMeasuresofEconomicValue,”OsgoodeHallLawReview30(1992):
737–70. Russell Korobkin, “The Endowment Effect and Legal Analysis,” Northwestern
UniversityLawReview97(2003):1227–93.
asymmetricaleffectsonindividualwell-being:Zamir,“LawandPsychology.”
29:TheFourfoldPattern
andotherdisasters:Includingexposuretoa“Dutchbook,”whichisasetofgamblesthat
yourincorrectpreferencescommityoutoacceptan{to>
puzzle that Allais constructed: Readers who are familiar with the Allais paradoxes will
recognizethatthisversionisnew.Itisbothmuchsimplerandactuallyastrongerviolation
than the original paradox. The left-hand option is preferred in the first problem. The
second problem is obtained by adding a more valuable prospect to the left than to the
right,buttheright-handoptionisnowpreferred.
sorely disappointed: As the distinguished economist Kenneth Arrow recently described
theevent,theparticipantsinthemeetingpaidlittleattentiontowhathecalled“Allais’s
littleexperiment.”Personalconversation,March16,2011.
estimatesforgains:Thetableshowsdecisionweightsforgains.Estimatesforlosseswere
verysimilar.
estimated from choices: Ming Hsu, Ian Krajbich, Chen Zhao, and Colin F. Camerer,
“Neural Response to Reward Anticipation under Risk Is Nonlinear in Probabilities,”
JournalofNeuroscience29(2009):2231–37.
parents of small children: W. Kip Viscusi, Wesley A. Magat, and Joel Huber, “An
InvestigationoftheRationalityofConsumerValuationsofMultipleHealthRisks,”RAND
JournalofEconomics18(1987):465–79.
psychologyofworry:Inarationalmodelwithdiminishingmarginalutility,peopleshould
payatleasttwo-thirdsasmuchtoreducethefrequencyofaccidentsfrom15to5unitsas
theyarewillingtopaytoeliminatetherisk.Observedpreferencesviolatedthisprediction.
not made much of it: C. Arthur Williams, “Attitudes Toward Speculative Risks as an
IndicatorofAttitudesTowardPureRisks,”JournalofRiskandInsurance33(1966):577–
86. Howard Raiffa, Decision Analysis: Introductory Lectures on Choices under
Uncertainty(Reading,MA:Addison-Wesley,1968).
shadowofciviltrials: Chris Guthrie, “Prospect Theory, Risk Preference, and the Law,”
NorthwesternUniversityLawReview97(2003):1115–63.JeffreyJ.Rachlinski,“Gains,
Losses and the Psychology of Litigation,” Southern California Law Review 70 (1996):
113–85.SamuelR.GrossandKentD.Syverud,“GettingtoNo:AStudyof Settlement
NegotiationsandtheSelectionofCasesforTrial,”MichiganLawReview90(1991):319–
93.
the frivolous claim: Chris Guthrie, “Framing Frivolous Litigation: A Psychological
Theory,”UniversityofChicagoLawReview67(2000):163–216.
30:RareEvents
wishtoavoidit:GeorgeF.Loewenstein,ElkeU.Weber,ChristopherK.Hsee,andNed
Welch,“RiskasFeelings,”PsychologicalBulletin127(2001):267–86.
vividness in decision making: Ibid. Cass R. Sunstein, “Probability Neglect: Emotions,
WorstCases,andLaw,”YaleLawJournal112(2002):61–107.Seenotestochapter13:
Damasio,Descartes’Error.Slovic,Finucane,Peters,andMacGregor,“The{r,n>:C.A
AffectHeuristic.”
Amos’s student: Craig R. Fox, “Strength of Evidence, Judged Probability, and Choice
UnderUncertainty,”CognitivePsychology38(1999):167–89.
focaleventandits:Judgmentsoftheprobabilitiesofaneventanditscomplementdonot
alwaysaddupto100%.Whenpeopleareaskedaboutatopictheyknowverylittleabout
(“WhatisyourprobabilitythatthetemperatureinBangkokwillexceed100°tomorrowat
noon?”), the judged probabilities of the event and its complement add up to less than
100%.
receiving a dozen roses: In cumulative prospect theory, decision weights for gains and
lossesarenotassumedtobeequal,astheywereintheoriginalversionofprospecttheory
thatIdescribe.
superficialprocessing:ThequestionaboutthetwournswasinventedbyDaleT.Miller,
WilliamTurnbull,andCathyMcFarland,“WhenaCoincidenceIsSuspicious:TheRoleof
Mental Simulation,” Journal of Personality and Social Psychology 57 (1989): 581–89.
Seymour Epstein and his colleagues argued for an interpretation of it in terms of two
systems: Lee A. Kirkpatrick and Seymour Epstein, “Cognitive-Experiential Self-Theory
and Subjective Probability: Evidence for Two Conceptual Systems,” Journal of
PersonalityandSocialPsychology63(1992):534–44.
judgedit as more dangerous:KimihikoYamagishi,“Whena12.86%MortalityIsMore
Dangerous Than 24.14%: Implications for Risk Communication,” Applied Cognitive
Psychology11(1997):495–506.
forensicpsychologists:Slovic,Monahan,andMacGregor,“ViolenceRiskAssessmentand
RiskCommunication.”
“1of1,000capitalcases”:JonathanJ.Koehler,“WhenArePeoplePersuadedbyDNA
MatchStatistics?”LawandHumanBehavior25(2001):493–513.
studiesofchoicefromexperience:RalphHertwig,GregBarron,ElkeU.Weber,andIdo
Erev, “Decisions from Experience and the Effect of Rare Events in Risky Choice,”
PsychologicalScience15(2004):534–39.RalphHertwigandIdoErev,“TheDescription-
ExperienceGapinRiskyChoice,”TrendsinCognitiveSciences13(2009):517–23.
notyetsettled:LiatHadarandCraigR.Fox,“InformationAsymmetryinDecisionfrom
DescriptionVersusDecisionfromExperience,”JudgmentandDecisionMaking4(2009):
317–25.
“chancesofrareevents”:HertwigandErev,“TheDescription-ExperienceGap.”
31:RiskPolicies
inferior option BC: The calculation is straightforward. Each of the two combinations
consistsofasurethingandagamble.Addthesurethingtobothoptionsofthegambleand
youwillfindADandBC.
the equivalent of “locking in”: Thomas Langer and Martin Weber, “Myopic Prospect
Theory vs. Myopic Loss Aversion: How General Is the Phenomenon?” Journal of E
{>Joenon?&conomicBehavior&Organization56(2005):25–38.
32:KeepingScore
driveintoablizzard:Theintuitionwasconfirmedinafieldexperimentinwhicharandom
selectionofstudentswhopurchasedseasonticketstotheuniversitytheaterreceivedtheir
ticketsatamuchreducedprice.Afollow-upofattendancerevealedthatstudentswhohad
paidthefullpricefortheirticketsweremorelikelytoattend,especiallyduringthefirst
halfoftheseason.Missingashowonehaspaidforinvolvestheunpleasantexperienceof
closinganaccountinthered.ArkesandBlumer,“ThePsychologyofSunkCosts.”
thedispositioneffect:HershShefrinandMeirStatman,“TheDispositiontoSellWinners
Too Early and Ride Losers Too Long: Theory and Evidence,” Journal of Finance 40
(1985): 777–90. Terrance Odean, “Are Investors Reluctant to Realize Their Losses?”
JournalofFinance53(1998):1775–98.
less susceptible: Ravi Dhar and Ning Zhu, “Up Close and Personal: Investor
SophisticationandtheDispositionEffect,”ManagementScience52(2006):726–40.
fallacycanbeovercome:DarrinR.Lehman,RichardO.Lempert,andRichardE.Nisbett,
“TheEffectsofGraduateTrainingonReasoning:FormalDisciplineandThinkingabout
Everyday-LifeEvents,”AmericanPsychologist43(1988):431–42.
“asinkingfeeling”:MarcelZeelenbergandRikPieters,“ATheoryofRegretRegulation
1.0,”JournalofConsumerPsychology17(2007):3–18.
regrettonormality:KahnemanandMiller,“NormTheory.”
habituallytakingunreasonable risks: Thehitchhikerquestionwasinspiredbya famous
example discussed by the legal philosophers Hart and Honoré: “A woman married to a
manwhosuffersfromanulceratedconditionofthestomachmightidentifyeatingparsnips
asthecause ofhisindigestion.Thedoctormightidentifytheulceratedconditionasthe
causeandthemealasamereoccasion.”Unusualeventscallforcausalexplanationsand
alsoevokecounterfactualthoughts,andthetwoarecloselyrelated.Thesameeventcanbe
compared to either a personal norm or the norm of other people, leading to different
counterfactuals, different causal attributions, and different emotions (regret or blame):
HerbertL.A.HartandTonyHonoré,CausationintheLaw(NewYork:OxfordUniversity
Press,1985),33.
remarkablyuniform:DanielKahnemanandAmosTversky,“TheSimulationHeuristic,”in
JudgmentUnderUncertainty:HeuristicsandBiases,ed.DanielKahneman,PaulSlovic,
andAmosTversky(NewYork:CambridgeUniversityPress,1982),160–73.
applies to blame: Janet Landman, “Regret and Elation Following Action and Inaction:
Affective Responses to Positive Versus Negative Outcomes,” Personality and Social
PsychologyBulletin13(1987):524–36.FaithGleicheretal.,“TheRoleofCounterfactual
ThinkinginJudgmentofAffect,”PersonalityandSocialPsychologyBulletin16(1990):
284–95.
actionsthatdeviatefromthedefault:DaleT.MillerandBrianR.Taylor,“Counterfactual
Thought,Regret,andSuperstition:HowtoAvoidKickingYourself,”inWhatMightHave
Been:TheSocialPsychologyofCounterfactualThinking,ed.NealJ.RoeseandJamesM.
Olson(Hillsdale,NJ:Erlbaum,1995),305–31.
produceblameandregret:MarcelZeelenberg,KeesvandenBos,EricvanDijk,andRik
Pieters, “The Inaction Effect in the Psychology of Regret,” Journal of Personality and
SocialPsychology82(2002):314–27.
brandnamesovergenerics:ItamarSimonson,“TheInfluenceofAnticipatingRegretand
ResponsibilityonPurchaseDecisions,”Journalof Consumer Research 19 (1992): 105–
18.
clean up their portfolios: Lilian Ng and Qinghai Wang, “Institutional Trading and the
Turn-of-the-YearEffect,”JournalofFinancialEconomics74(2004):343–66.
lossaverseforaspectsofyourlife:TverskyandKahneman,“LossAversioninRiskless
Choice.”EricJ.Johnson,SimonGächter,andAndreasHerrmann,“ExploringtheNature
ofLossAversion,”CentreforDecisionResearchandExperimentalEconomics,University
ofNottingham,DiscussionPaperSeries,2006.EdwardJ.McCaffery,DanielKahneman,
and Matthew L. Spitzer, “Framing the Jury: Cognitive Perspectives on Pain and
Suffering,”VirginiaLawReview81(1995):1341–420.
classic on consumer behavior: Richard H. Thaler, “Toward a Positive Theory of
ConsumerChoice,”JournalofEconomicBehaviorandOrganization39(1980):36–90.
tabootradeoff:PhilipE.Tetlocketal.,“ThePsychologyoftheUnthinkable:TabooTrade-
Offs,ForbiddenBaseRates,andHereticalCounterfactuals,”Journal of Personalityand
SocialPsychology78(2000):853–70.
where the precautionary principle: Cass R. Sunstein, The Laws of Fear: Beyond the
PrecautionaryPrinciple(NewYork:CambridgeUniversityPress,2005).
“psychologicalimmunesystem”:DanielT.Gilbertetal.,“LookingForwardtoLooking
Backward:TheMispredictionofRegret,”PsychologicalScience15(2004):346–50.
33:Reversals
intheman’sregularstore:DaleT.MillerandCathyMcFarland,“CounterfactualThinking
andVictimCompensation:ATestofNormTheory,”PersonalityandSocialPsychology
Bulletin12(1986):513–19.
reversals of judgment and choice: The first step toward the current interpretation was
takenbyMaxH. Bazerman, GeorgeF. Loewenstein,andSallyB.White,“Reversals of
Preference in Allocation Decisions: Judging Alternatives Versus Judging Among
Alternatives,” Administrative Science Quarterly 37 (1992): 220–40. Christopher Hsee
introducedtheterminologyofjointandseparateevaluation,andformulatedtheimportant
evaluability hypothesis, which explains reversals by the idea that some attributes {e
a#822becomeevaluableonlyinjointevaluation:“AttributeEvaluability:ItsImplications
forJoint-SeparateEvaluationReversalsandBeyond,”inKahnemanandTversky,Choices,
Values,andFrames.
conversationbetweenpsychologistsandeconomists:SarahLichtensteinandPaulSlovic,
“ReversalsofPreferenceBetweenBidsandChoicesinGamblingDecisions,”Journalof
ExperimentalPsychology89(1971):46–55.Asimilarresultwasobtainedindependently
by Harold R. Lindman, “Inconsistent Preferences Among Gambles,” Journal of
ExperimentalPsychology89(1971):390–97.
bewilderedparticipant: For a transcript of the famous interview, see Sarah Lichtenstein
andPaulSlovic,eds.,TheConstructionofPreference(NewYork:CambridgeUniversity
Press,2006).
the prestigious American Economic Review: David M. Grether and Charles R. Plott,
“Economic Theory of Choice and the Preference Reversals Phenomenon,” American
EconomicReview69(1979):623–28.
“context in which the choices are made”: LichtensteinandSlovic, The Construction of
Preference,96.
one embarrassing finding: Kuhn famously argued that the same is true of physical
sciencesas well:ThomasS. Kuhn,“The Functionof Measurement inModern Physical
Science,”Isis52(1961):161–93.
likingofdolphins:Thereisevidencethatquestionsabouttheemotionalappealofspecies
and the willingness to contribute to their protection yield the same rankings: Daniel
KahnemanandIlanaRitov,“DeterminantsofStatedWillingnesstoPayforPublicGoods:
AStudyintheHeadlineMethod,”JournalofRiskandUncertainty9(1994):5–38.
superioronthisattribute:Hsee,“AttributeEvaluability.”
“requisite record-keeping”: Cass R. Sunstein, Daniel Kahneman, David Schkade, and
IlanaRitov,“PredictablyIncoherentJudgments,”StanfordLawReview54(2002):1190.
34:FramesandReality
unjustifiedinfluencesofformulation:AmosTverskyandDanielKahneman,“TheFraming
ofDecisionsandthePsychologyofChoice,”Science211(1981):453–58.
paidwithcashoroncredit:Thaler,“TowardaPositiveTheoryofConsumerChoice.”
10%mortalityisfrightening:BarbaraMcNeil,StephenG.Pauker,HaroldC.SoxJr.,and
Amos Tversky, “On the Elicitation of Preferences for Alternative Therapies,” New
EnglandJournalofMedicine306(1982):1259–62.
“Asian disease problem”: Some people have commented that the “Asian” label is
unnecessary and pejorative. We probably would not use it today, but the example was
writteninthe1970s,whensensitivitytogrouplabelswaslessdevelopedthanitistoday.
Thewordwasaddedtomaketheexamplemoreconcretebyremindingrespondentsofthe
Asianfluepidem{anslessicof1957.
ChoiceandConsequence:ThomasSchelling,ChoiceandConsequence(Cambridge,MA:
HarvardUniversityPress,1985).
misleadingframe:RichardP.LarrickandJackB.Soll,“TheMPGIllusion,”Science320
(2008):1593–94.
rateoforgandonationinEuropeancountries:EricJ.JohnsonandDanielGoldstein,“Do
DefaultsSaveLives?”Science302(2003):1338–39.
35:TwoSelves
“wantability”: Irving Fisher, “Is ‘Utility’ the Most Suitable Term for the Concept It Is
UsedtoDenote?”AmericanEconomicReview8(1918):335.
atanymoment:FrancisEdgeworth,MathematicalPsychics(NewYork:Kelley,1881).
under which his theory holds: Daniel Kahneman, Peter P. Wakker, and Rakesh Sarin,
“BacktoBentham?ExplorationsofExperiencedUtility,”QuarterlyJournalofEconomics
112(1997):375–405.DanielKahneman,“ExperiencedUtilityandObjectiveHappiness:
A Moment-Based Approach” and “Evaluation by Moments: Past and Future,” in
KahnemanandTversky,Choices,Values,andFrames,673–92,693–708.
a physician and researcher: Donald A. Redelmeier and Daniel Kahneman, “Patients’
Memories of Painful Medical Treatments: Real-time and Retrospective Evaluations of
TwoMinimallyInvasiveProcedures,”Pain66(1996):3–8.
free to choose: Daniel Kahneman, Barbara L. Frederickson, Charles A. Schreiber, and
Donald A. Redelmeier, “When More Pain Is Preferred to Less: Adding a Better End,”
PsychologicalScience4(1993):401–405.
duration of the shock: Orval H. Mowrer and L. N. Solomon, “Contiguity vs. Drive-
Reduction in Conditioned Fear: The Proximity and Abruptness of Drive Reduction,”
AmericanJournalofPsychology67(1954):15–25.
burstofstimulation:PeterShizgal,“OntheNeuralComputationofUtility:Implications
fromStudiesofBrainStimulationReward,”inWell-Being:TheFoundationsofHedonic
Psychology, ed. Daniel Kahneman, Edward Diener, and Norbert Schwarz (New York:
RussellSageFoundation,1999),500–24.
36:LifeasaStory
hadalover: Paul Rozinand JenniferStellar, “Posthumous EventsAffectRatedQuality
andHappinessofLives,”JudgmentandDecisionMaking4(2009):273–79.
entirelivesaswellasbriefepisodes:EdDiener,DerrickWirtz,andShigehiroOishi,“End
EffectsofRatedLifeQuality:TheJamesDeanEffect,”PsychologicalScience12(2001):
124–28.Thesameseriesofexperimentsalsotestedforthepeak-endruleinanunhappy
lifeandfoundsimilarresults:Jenwasnotjudgedtwiceasunhappyifshelivedmiserably
for 60 years rather than 30, but { thk-e she was regarded as considerably happier if 5
mildlymiserableyearswereaddedjustbeforeherdeath.
37:ExperiencedWell-Being
lifeasawholethesedays:Anotherquestionthathasbeenusedfrequentlyis,“Takenall
together, how would you say things are these days? Would you say that you are very
happy,prettyhappy,or nottoohappy?”ThisquestionisincludedintheGeneralSocial
Survey in the United States, and its correlations with other variables suggest a mix of
satisfaction and experienced happiness. A pure measure of life evaluation used in the
GallupsurveysistheCantrilSelf-AnchoringStrivingScale,inwhichtherespondentrates
hisorhercurrentlifeonaladderscaleinwhich0is“theworstpossiblelifeforyou”and
10is“thebestpossiblelifeforyou.”Thelanguagesuggeststhatpeopleshouldanchoron
what they consider possible for them, but the evidence shows that people all over the
world have a common standard for what a good life is, which accounts for the
extraordinarilyhighcorrelation(r=.84)betweentheGDPofcountriesandtheaverage
ladderscoreoftheircitizens.AngusDeaton,“Income,Health,andWell-BeingAroundthe
World: Evidence from the Gallup World Poll,” Journal of Economic Perspectives 22
(2008):53–72.
“adreamteam”:TheeconomistwasAlanKruegerofPrinceton,notedforhisinnovative
analyses of unusual data. The psychologists were David Schkade, who had
methodological expertise; Arthur Stone, an expert on health psychology, experience
sampling,andecologicalmomentaryassessment;NorbertSchwarz,asocialpsychologist
whowasalsoanexpertonsurveymethodandhadcontributedexperimentalcritiquesof
well-beingresearch,includingtheexperimentonwhichadimeleftonacopyingmachine
influencedsubsequentreportsoflifesatisfaction.
intensity of various feelings: In some applications, the individual also provides
physiologicalinformation,suchascontinuousrecordingsofheartrate,occasionalrecords
of blood pressure, or samples of saliva for chemical analysis. The method is called
Ecological Momentary Assessment: Arthur A. Stone, Saul S. Shiffman, and Marten W.
DeVries, “Ecological Momentary Assessment Well-Being: The Foundations of Hedonic
Psychology,”inKahneman,Diener,andSchwarz,Well-Being,26–39.
spend their time: Daniel Kahneman et al., “A Survey Method for Characterizing Daily
LifeExperience:TheDayReconstructionMethod,”Science306(2004):1776–80.Daniel
KahnemanandAlanB.Krueger,“DevelopmentsintheMeasurementofSubjectiveWell-
Being,”JournalofEconomicPerspectives20(2006):3–24.
physiologicalindicationsof emotion:Previousresearchhad documented thatpeopleare
able to “relive” feelings they had in a past situation when the situation is retrieved in
sufficientlyvividdetail.MichaelD.RobinsonandGeraldL.Clore,“BeliefandFeeling:
Evidence for an Accessibility Model of Emotional Self-Report,” Psychological Bulletin
128(2002):934–60.
statetheU-index:AlanB.Krueger,ed.,MeasuringtheSubjectiveWell-BeingofNations:
National Accounts of Time Use and Well-Being (Chicago: University of Chicago Press,
2009).
distributio{i>dll-Being:EdDiener,“MostPeopleAreHappy,”PsychologicalScience7
(1996):181–85.
Gallup World Poll: For a number of years I have been one of several Senior Scientists
associatedwiththeeffortsoftheGallupOrganizationinthedomainofwell-being.
more than 450,000 responses: Daniel Kahneman and Angus Deaton, “High Income
ImprovesEvaluationofLifebutNotEmotionalWell-Being,”ProceedingsoftheNational
AcademyofSciences107(2010):16489–93.
worsefortheverypoor:DylanM.Smith,KennethM.Langa,MohammedU.Kabeto,and
PeterUbel,“Health,Wealth,andHappiness:FinancialResourcesBufferSubjectiveWell-
BeingAftertheOnsetofaDisability,”PsychologicalScience16(2005):663–66.
$75,000 in high-cost areas: In a TED talk I presented in February 2010 I mentioned a
preliminaryestimateof$60,000,whichwaslatercorrected.
eatabarofchocolate!:JordiQuoidbach,ElizabethW.Dunn,K.V.Petrides,andMoïra
Mikolajczak, “Money Giveth, Money Taketh Away: The Dual Effect of Wealth on
Happiness,”PsychologicalScience21(2010):759–63.
38:ThinkingAboutLife
German Socio-Economic Panel: Andrew E. Clark, Ed Diener, and Yannis Georgellis,
“LagsandLeadsinLifeSatisfaction:ATestoftheBaselineHypothesis.”Paperpresented
attheGermanSocio-EconomicPanelConference,Berlin,Germany,2001.
affectiveforecasting:DanielT.GilbertandTimothyD.Wilson,“WhytheBrainTalksto
Itself:SourcesofErrorinEmotionalPrediction,”PhilosophicalTransactionsoftheRoyal
SocietyB364(2009):1335–41.
only significant fact in their life: Strack, Martin, and Schwarz, “Priming and
Communication.”
questionnaireonlifesatisfaction:TheoriginalstudywasreportedbyNorbertSchwarzin
hisdoctoralthesis (in German)“Moodas Information: OntheImpactof Moodsonthe
EvaluationofOne’sLife”(Heidelberg:SpringerVerlag,1987).Ithasbeendescribedin
many places, notably Norbert Schwarz and Fritz Strack, “Reports of Subjective Well-
Being: Judgmental Processes and Their Methodological Implications,” in Kahneman,
Diener,andSchwarz,Well-Being,61–84.
goals that young people set: The study was described in William G. Bowen and Derek
Curtis Bok, The Shape of the River: Long-Term Consequences of Considering Race in
CollegeandUniversityAdmissions(Princeton:Princeton University Press, 1998). Some
ofBowenandBok’sfindingswerereportedbyCarolNickerson,NorbertSchwarz,andEd
Diener, “Financial Aspirations, Financial Success, and Overall Life Satisfaction: Who?
andHow?”JournalofHappinessStudies8(2007):467–515.
“being very well-off financially”: Alexander Astin, M. R. King, and G. T. Richardson,
“The American Freshman: National Norms for Fall 1976,” Cooperative Institutional
ResearchProgramoftheAmericanC{heon,RouncilonEducationandtheUniversityof
California at Los Angeles, Graduate School of Education, Laboratory for Research in
HigherEducation,1976.
money was not important: These results were presented in a talk at the American
Economic Association annual meeting in 2004. Daniel Kahneman, “Puzzles of Well-
Being,”paperpresentedatthemeeting.
happiness of Californians: The question of how well people today can forecast the
feelingsoftheirdescendantsahundredyearsfromnowisclearlyrelevanttothepolicy
response to climate change, but it can be studied only indirectly, which is what we
proposedtodo.
aspectsoftheirlives:Inposingthequestion,IwasguiltyofaconfusionthatInowtryto
avoid:Happinessandlifesatisfactionarenotsynonymous.Lifesatisfactionreferstoyour
thoughts and feelings when you think about your life, which happens occasionally—
includinginsurveysofwell-being.Happinessdescribesthefeelingspeoplehaveasthey
livetheirnormallife.
Ihadwonthefamilyargument:However,mywifehasneverconceded.Sheclaimsthat
onlyresidentsofNorthernCaliforniaarehappier.
students in California and in the Midwest: Asian students generally reported lower
satisfactionwiththeirlives,andAsianstudentsmadeupamuchlargerproportionofthe
samplesinCaliforniathanintheMidwest.Allowingforthisdifference,lifesatisfactionin
thetworegionswasidentical.
Howmuchpleasuredoyougetfromyourcar?:JingXuandNorbertSchwarzhavefound
thatthequalityofthecar(asmeasuredbyBlueBookvalue)predictstheowners’answer
toageneralquestionabouttheirenjoymentofthecar,andalsopredictspeople’spleasure
duringjoyrides.Butthequalityofthecarhasnoeffectonpeople’smoodduringnormal
commutes. Norbert Schwarz, Daniel Kahneman, and Jing Xu, “Global and Episodic
Reports of Hedonic Experience,” in R. Belli, D. Alwin, and F. Stafford (eds.), Using
Calendar and Diary Methods in Life Events Research (Newbury Park, CA: Sage), pp.
157–74.
paraplegicsspendinabadmood?:ThestudyisdescribedinmoredetailinKahneman,
“EvaluationbyMoments.”
think about their situation: Camille Wortman and Roxane C. Silver, “Coping with
IrrevocableLoss,Cataclysms,Crises,andCatastrophes:PsychologyinAction,”American
PsychologicalAssociation,MasterLectureSeries6(1987):189–235.
studiesofcolostomypatients:DylanSmithetal.,“MisrememberingColostomies?Former
Patients Give Lower Utility Ratings than Do Current Patients,” Health Psychology 25
(2006): 688–95. George Loewenstein and Peter A. Ubel, “Hedonic Adaptation and the
RoleofDecisionandExperienceUtilityinPublicPolicy,”JournalofPublicEconomics
92(2008):1795–1810.
the word miswanting: Daniel Gilbert and Timothy D. Wilson, “Miswanting: Some
ProblemsinAffectiveForecasting,”inFeelingandThinking:TheRoleofAffectinSocial
Cognition,ed.JosephP.Forgas(NewYork:CambridgeUniversityPress,2000),178–97.
Conclusions
tooimportanttobeignored:PaulDolanandDanielKahneman,“InterpretationsofUtility
andTheirImplicationsfortheValuationofHealth,”EconomicJournal118(2008):215–
234. Loewenstein and Ubel, “Hedonic Adaptation and the Role of Decision and
ExperienceUtilityinPublicPolicy.”
guidegovernmentpolicies:ProgresshasbeenespeciallyrapidintheUK,wheretheuseof
measures of well-being is now official government policy. These advances were due in
goodparttotheinfluenceofLordRichardLayard’sbookHappiness:LessonsfromaNew
Science, first published in 2005. Layard is among the prominent economists and social
scientistswho have been drawn into thestudyofwell-beinganditsimplications.Other
important sources are: Derek Bok, The Politics of Happiness: What Government Can
Learn from the New Research on Well-Being (Princeton: Princeton University Press,
2010).EdDiener,RichardLucus,UlrichSchmimmack,andJohnF.Helliwell,Well-Being
for Public Policy (New York: Oxford University Press, 2009). Alan B. Krueger, ed.,
MeasuringtheSubjectiveWell-BeingofNations:NationalAccountofTimeUseandWell-
Being(Chicago:UniversityofChicagoPress,2009).JosephE.Stiglitz,AmartyaSen,and
Jean-Paul Fitoussi, Report of the Commission on the Measurement of Economic
Performance and Social Progress. Paul Dolan, Richard Layard, and Robert Metcalfe,
Measuring Subjective Well-being for Public Policy: Recommendations on Measures
(London:OfficeforNationalStatistics,2011).
Irrational is a strong word: The view of the mind that Dan Ariely has presented in
PredictablyIrrational:TheHiddenForcesThatShapeOurDecisions(NewYork:Harper,
2008)isnotmuchdifferentfrommine,butwedifferinouruseoftheterm.
accept future addiction: Gary S. Becker and Kevin M. Murphy, “A Theory of Rational
Addiction,” Journal of Political Economics 96 (1988): 675–700. Nudge: Richard H.
Thaler and Cass R. Sunstein, Nudge: Improving Decisions About Health, Wealth, and
Happiness(NewHaven:YaleUniversityPress,2008).
can institute and enforce: Atul Gawande, The Checklist Manifesto: How to Get Things
Right(NewYork:Holt,2009).DanielKahneman,DanLovallo,andOliverSibony,“The
BigIdea:BeforeYouMakeThatBigDecision…”HarvardBusinessReview89 (2011):
50–60.
distinctivevocabulary:ChipHeath,RichardP.Larrick,andJoshuaKlayman,“Cognitive
Repairs: How Organizational Practices Can Compensate for Individual Shortcomings,”
ResearchinOrganizationalBehavior20(1998):1–37.
P
Index
Theindexthatappearedintheprintversionofthistitledoesnotmatchthepagesinyour
eBook. Please use the search function on your eReading device to search for terms of
interest.Foryourreference,thetermsthatappearintheprintindexarelistedbelow.
adaptationlevel
Add-1task
adjustment;insufficient
affectheuristic;availabilityand
affectiveforecasting
airplanecrashes
Ajzen,Icek
Alarscare
algorithms;Apgarscores;hostilityto;multipleregression
Allais,Maurice
al-Qaeda
ambiguity,suppressionof
AmericanEconomicReview
amygdala
anchoringindex
anchors, anchoring; as adjustment; associative coherence in; associative memory and;
measurement of; as priming effect; random, power of; in System 1 and System 2;
usesandabusesof
anesthesiologists
angryfaces
anomalies
anteriorcingulate
Apgar,Virginia
Apgarscores
aphorisms
Ariely,Dan
Arrow,Kenneth
artexperts
artifacts,inresearch
Asch,Solomon
Ashenfelter,Orley
Asiandiseaseproblem
assessments,basic
associations;activatedideasin;causalityand;primingand
associative coherence; in anchoring; halo effect and; plausibility and, associative
coherence(cont.);WYSIATI(whatyouseeisallthereis)and
associative memory; abnormal events and; anchoring and; causality and; confirmation
biasand;creativityand;andestimatesofcausesofdeath
Åstebro,Thomas
Atlantic,The
attention;inself-control
paneight=“0%”width=”-5%”>
AttentionandEffort(Kahneman)
Auerbach,Red
authoritarianideas
availability; affect and; and awareness of one’s biases; expectations about; media and;
psychologyof;riskassessmentand,seeriskassessment
availabilitycascades
availabilityentrepreneurs
badandgood,distinctionsbetween
banks
banktellerproblem
Barber,Brad
Bargh,John
baseball
baseballcards
baselinepredictions
baserates;incabdriverproblem;causal;inhelpingexperiment;low;statistical;inTom
Wproblem;inYaleexamproblem
basicassessments
basketball
basketballtickets
bat-and-ballproblem
Baumeister,Roy
Bayes,Thomas
Bayesianstatistics
Bazerman,Max
Beane,Billy
Beatty,Jackson
Becker,Gary
“BecomingFamousOvernight”(Jacoby)
behavioraleconomics
BehavioralInsightTeam
“BeliefintheLawofSmallNumbers”(TverskyandKahneman)
beliefs:biasfor;past,reconstructionof
Benartzi,Shlomo
Bentham,Jeremy
Berlin,Isaiah
Bernoulli,Daniel
Bernouilli,Nicholas
Beyth,Ruth
bicyclemessengers
BlackSwan,The(Taleb)
blame
Blink(Gladwell)
Borg,Björn
Borgida,Eugene
“BoysWillBeBoys”(BarberandOdean)
Bradlee,Ben
brain;amygdalain;anteriorcingulatein;buyingandsellingand;emotionalframing
and;frontalareaof;pleasureand;prefrontalareaof;punishmentand;sugarin;
threatsand;andvariationsofprobabilities
BritishToxicologySociety
broadframing
Brockman,John
broken-legrule
budgetforecasts
BuilttoLast(CollinsandPorras)
Bush,GeorgeW.
businessandleadershippractices;atGoogle
businesspundits
Cabanac,Michel
cabdriverproblem
cabdrivers,NewYorkCity
Californians
Camerer,Colin
cancer;surgeryvs.radiationfor
CantrilSelf-AnchoringStrivingScale
Carroll,Lewis
carsanddriving;brakesin;drivingtests;fueleconomyand;pleasurefrom
cashbox
categories
causalbaserates
causalinterpretations;correlationand;regressioneffectsand
causalsituations
causalstereotypes
causes,andstatistics
CEOs;optimistic
certaintyeffect
CFOs
Chabris,Christopher
chanceandrandomness;misconceptionsof
changingone’smind
ChecklistManifesto,A(Gawande)
chess
children:caringfor;depressed;timespentwith
China
ChoiceandConsequence(Schelling)
choicearchitecture
choices: from description; from experience; see also decisions, decision
making;riskassessment
“Choices,Values,andFrames”(KahnemanandTversky)
CIA
Clark,Andrew
climate
Clinicalvs.StatisticalPrediction:ATheoreticalAnalysisandaReviewofthe
Evidence(Meehl)
Clinton,Bill
Coelho,Marta
coffeemugexperiments
cognitivebusyness
cognitive ease; in basic assessments; and illusions of remembering; and
illusionsoftruth;moodand;andwritingpersuasivemessages;WYSIATI
(whatyouseeisallthereis)and
cognitive illusions; confusing experiences with memories; of pundits; of
remembering;ofskill;ofstock-pickingskill;oftruth;ofunderstanding;of
validity
CognitiveReflectionTest(CRT)
cognitivestrain
Cohen,David
coherence;seealsoassociativecoherence
Cohn,Beruria
coincidence
coin-on-the-machineexperiment
cold-handexperiment
Collins,Jim
colonoscopies
colostomypatients
competence,judgingof
competitionneglect
complexvs.simplelanguage
concentration
cogndivheight=“0%”>
“ConditionsforIntuitiveExpertise:AFailuretoDisagree”(KahnemanandKlein)
confidence;biasof,overdoubt;overconfidence;WYSIATI(whatyouseeisallthereis)
and
confirmationbias
conjunctionfallacy
conjunctiveevents,evaluationof
“ConsequencesofEruditeVernacularUtilizedIrrespectiveofNecessity:Problemswith
UsingLongWordsNeedlessly”(Oppenheimer)
contiguityintimeandplace
control
cookieexperiment
correlation;causationand;illusory;regressionand;sharedfactorsand
correlationcoefficient
cost-benefitcorrelation
costs
creativity;associativememoryand
credibility
Csikszentmihalyi,Mihaly
curriculumteam
Damasio,Antonio
datingquestion
Dawes,Robyn
DayReconstructionMethod(DRM)
death:causesof;lifestoriesand;organdonationand;remindersof
Deaton,Angus
decisions, decision making; broad framing in; and choice from description; and choice
fromexperience;emotionsandvividnessin;expectationprinciplein;ingambles,see
gambles;globalimpressionsand;hindsightbiasand;narrowframingin;optimistic
biasin;planningfallacyand;povertyand;premortemand;referencepointsin;regret
and;riskand,seeriskassessment
decisionutility
decision weights; overweighting; unlikely events and; in utility theory vs. prospect
theory;vividoutcomesand;vividprobabilitiesand
decorrelatederrors
defaultoptions
denominatorneglect
depression
Detroit/Michiganproblem
Diener,Ed
dierollproblem
dinnerwareproblem
disclosures
diseasethreats
disgust
disjunctiveevents,evaluationof
dispositioneffect
DNAevidence
dolphins
Dosi,Giovanni
doubt;biasofconfidenceover;premortemand;suppressionof
DukeUniversity
Duluth,Minn.,bridgein
durationneglect
durationweighting
earthquakes
eating
eBay
Econometrica
economics; behavioral; Chicago school of; neuroeconomics; preference reversals
and;rational-agentmodelin
economictransactions,fairnessin
EconsandHumans
Edge
Edgeworth,Francis
education
effectivenessofsearchsets
effort;least,lawof;inself-control
egodepletion
electricity
electricshocks
emotionalcoherence,seehaloeffectemotionallearning
emotionsandmood:activitiesand;affectheuristic;availabilitybiasesand;inbasic
assessments;cognitiveeaseand;indecisionmaking;inframing;moodheuristic
for happiness; negative, measuring; and outcomes produced by action vs.
inaction; paraplegics and; perception of; substitution of question on; in vivid
outcomes;invividprobabilities;weatherand;workand
employers,fairnessrulesand
endangeredspecies
endowmenteffect;andthinkinglikeatrader
energy,mental
engagement
EnquiryConcerningHumanUnderstanding,An(Hume)
entrepreneurs;competitionneglectby
Epley,Nick
Epstein,Seymour
equal-weightingschemes
Erev,Ido
evaluabilityhypothesis
evaluations:joint;jointvs.single;single
evidence:one-sided;ofwitnesses
executivecontrol
expectationprinciple
expectations
expectedutilitytheory,seeutilitytheory
experiencedutility
experiencesampling
experiencingself;well-beingof;seealsowell-being
expert intuition; evaluating; illusions of validity of; overconfidence and; as
recognition;riskassessmentand;vs.statisticalpredictions;trustin
expertise,seeskill
ExpertPoliticalJudgment:HowGoodIsIt?HowCanWeKnow?(Tetlock)
ExxonValdezoilspill
eyes,pupildilationin
facereading
fairness
fallacies;conjunction;narrative;planning;sunk-cost
familiarity
FarSide,The(Larson)
fastandfrugalheuristic
fastthinking
fatigue
fear
Fechner,Gustav
feedback
Feller,William
financialcrisisof2008
fi
nancialadvisersandforecasters
firefighters
firstimpressions
Fischhoff,Baruch
flightinstructors
floodmonitor
Floridaeffect
flow
flowerssyllogism
Flyvbjerg,Bent
focus
focusingillusion
fonts
forecasts,seepredictionsandforecasts
footballgame
FordMotorCompany
formulas;algorithms;Apgarscores;hostilityto;forinterviews;multipleregression
formulationeffects
Fortune
fourfoldpattern;inlegalcases
Fox,Craig
Fox,Seymour
frames, framing; in Asian disease problem; in child exemption problem; in
disclosures;emotional;fuel economy and; good; inKEEP-LOSEstudy; organ
donationand;regulationson;insurvival-mortalityexperiment;inticketproblem
Frederick,Shane
Freedman,David
freedom
FreetoChoose(Friedman)
frequencyrepresentation
Frey,Bruno
Friedman,Milton
frowning;availabilityheuristicand;representativenessand
gains
Galinsky,Adam
Gallup-HealthwaysWell-BeingIndex
Galton,Francis
gambles;bundlingof;certaintyeffectand;emotionalframingin;lossaversionin;
lottery;mixed;andoutcomesproducedbyactionvs.inaction;possibilityeffect
and;psychologicalvalueof;regretand;simple;St.Petersburgparadoxand;vs.
surethings;utilityongsv>seealsoriskassessment
GatesFoundation
Gawande,Atul
Georgellis,Yannis
GermanSocio-EconomicPanel
gestures
Gibbs,Lois
Gigerenzer,Gerd
Gilbert,Daniel
Gilovich,Tom
Gladwell,Malcolm
globalwarming
glucose
goals
golf
goodandbad,distinctionsbetween
Google
gorillaexperiment
gossip
Gottman,John
Gould,StephenJay
gradesandgradepointaverages(GPAs)
gradingstudents’essays
Grether,David
group,joining
Guthrie,Chris
Haidt,Jonathan
haloeffect
HaloEffect,The(Rosenzweig)
happiness; of Californians; dating question and; income and; life stories and;
marriageand;moodheuristicfor;seealsowell-beinghappyfaces
happywords
Harding,WarrenG.
HarvardMedicalSchool
HarvardUniversity
health:diseasethreatsand;well-beingand;risksand;seealsomedicine
healthsurveyproblem
healthviolationpenalties
HebrewUniversityofJerusalem
“HedgehogandtheFox,The”(Berlin)
hedonimeter
Heider,Fritz
helpingexperiment
Hertwig,Ralph
Hess,Eckhard
heuristic,definitionof
highschoolcurriculumteam
hindsight:biasin;regretand
historicalevents
hitchhikerquestion
Hitler,Adolf
Hogarth,Robin
honestybox
“HowMentalSystemsBelieve”(Gilbert)
HowtoSolveIt(Pólya)
Hsee,Christopher
hubrishypothesis
HumansandEcons
Hume,David
hunger
hypotheses,testing
ideomotoreffect
illusions:cognitive,seecognitiveillusions;Müller-Lyer;3-D
imaginability,immediategratification
incongruity
independentjudgments
indifferencemap
inheritances
injectionpuzzle
InSearchofExcellence(PetersandWaterman)
insideview
insurance
intelligence;inmarriage;pretentiouslanguageand
intensitymatching
intention
interviews;inIsraeliDefenseForces
IntroductiontothePrinciplesofMoralsandLegislation(Bentham)
intuition: acquisiitiodution of; common use of word; of experts, see expert intuition;
predictive,seepredictionsandforecasts;asrecognition;Simon’sdefinitionof
Inventor’sAssistanceProgram
investments:stockportfolios;sunk-costfallacyand
InvisibleGorilla,The(ChabrisandSimons)
irrationality
Israel,bombingsin
IsraeliDefenseForces:flightinstructorsin;interviewsin;leaderlessgroupchallengein
IsraeliMinistryofEducation
“Jabberwocky”(Carroll)
Jacoby,Larry
Jencks,Christopher
jointevaluations;singleevaluationsvs.
judgmentheuristics
JudgmentinManagerialDecisionMaking(Bazerman)
judgments;basicassessmentsin;ofexperts,seeexpertintuition;intensitymatchingin;
mentalshotgunin;predictive,seepredictionsandforecasts;setsandprototypesin;
summary,ofcomplexinformation;seealsodecisions,decisionmaking
“JudgmentUnderUncertainty:HeuristicsandBiases”(TverskyandKahneman)
Julieproblem
jumpingtoconclusions;biasforbeliefandconfirmationin;haloeffectin,seehaloeffect;
suppressionofambiguityanddoubtin;WYSIATIin,seewhatyouseeisallthereis
Kaye,Danny
keepingscore;mentalaccountsand;regretand;responsibilityand
KEEP-LOSEstudy
kidneycancer
KillingGround,The
kitchenrenovations
Klein,Gary
Knetsch,Jack
know,useofword
knowledge;reconstructionofpaststatesof
kouros
Krueger,Alan
Kunreuther,Howard
Kuran,Timur
labornegotiations
LadyMacbetheffect
language,complexvs.simple
Larrick,Richard
Larson,Gary
law,seelegalcaseslawoflargenumbers
lawofsmallnumbers;andbiasofconfidenceoverdoubt
lazinessofSystem2
Layard,Richard
leaderlessgroupchallenge
leadershipandbusinesspractices;atGoogle
LeBoeuf,Robyn
legal cases: civil, damages in; DNA evidence in; fourfold pattern and; frivolous; loss
aversionin;malpractice;outcomebiasin
leisuretime
less-is-morepattern
Lewis,Michael
libertarianpolicies
Lichtenstein,Sarah
life:evaluationof;storiesin;satisfactionin;thinkingabout
Lindaproblem
List,John
loans
logarithmicfunctions
lossaversion;inanimals;enhanced;goalsasreferencepointsin;inlegaldecisions;status
quoand
lossaversionratio
losses
lotteries
Lovallo,Dan
LoveCanal
luck
lying
Malkiel,Burton
Malmendier,Ulrike
malpracticelitigation
MaoZedong
marchofhistoryuote>
Markowitz,Harry
marriage;lifesatisfactionand
MathematicalPsychology(Dawes,Tversky,andCoombs)
matter,relationofmindto
McFarland,Cathy
media,availabilityheuristicand
medicalschooladmissions
medicalsurveyproblem
medicine;expertisein;malpracticelitigation;overconfidencein;physicians;unique
casesin;unusualtreatmentsin
Mednick,Sarnoff
Meehl,Paul
meetings
memory,memories;associative,seeassociativememory;availabilityheuristicand,
seeavailability;durationneglectin;experiencedutilityand;illusionsof;andthe
rememberingself;ofvacations
mentalaccounts
mentaleffort,seeeffortmentalenergy
mentalshotgun
mereexposureeffect
messages,persuasive
metaphors
Michigan/Detroitproblem
MichiganStateUniversity
Michotte,Albert
Miller,Dale
mind,relationofmatterto
Mischel,Walter
miswanting
MIT
moneyandwealth:culturaldifferencesinattitudestoward;happinessand;income
vs.leisure;mentalaccountsand;poverty;primingand;utilityof
Moneyball(Lewis)
mood,seeemotionsandmoodMorgenstern,Oskar
Mosesillusion
motivation
movies
“MPGIllusion,The”(LarrickandSoll)
mugexperiments
Mullainathan,Sendhil
Müller-Lyerillusion
multipleregression
Mussweiler,Thomas
mutualfunds
names:complicated;offamouspeople
narrativefallacy
narrowframing;dispositioneffect
NaturalisticDecisionMaking(NDM)
negativitydominance
negotiations
neuroeconomics
NewYorkTimes,The
NewYorkUniversity
9/11
Nisbett,Richard
Nixon,Richard
NobelPrize
norms
normtheory
novelty
Nudge(ThalerandSunstein)
nutrition
OaklandA’s
Obama,Barack
obesity
Odean,Terry
OfficeofInformationandRegulatoryAffairs
one-sidedevidence
Oppenheimer,Danny
optimalexperience
optimism;inCEOs;resilienceand
optimistic bias; competition neglect; in entrepreneurs; overconfidence; planning
fallacy;premortemand;risktakingand
OregonResearchInstitute
organdonation
organizations
outcomebias
outsideview
ou>
pain; chronic; cold-hand experiment and; colonoscopies and; duration neglect
and;injectionpuzzleand;memoryof;operationexperimentand;peak-endrule
and;inrats
paraplegics
parole
past:andconfusingexperienceswithmemories;hindsightbiasand;regretand
pastness
patternseeking
Pavlov,Ivan
peak-endrule
persuasivemessages
physicians;malpracticelitigationand
pianoplayingandweight,measuring
planecrashes
planningfallacy;mitigating
plausibility
pleasure;inrats
Plott,Charles
poignancy
politicalexperts
politicalpreference
Pólya,George
Pope,Devin
Porras,JerryI.
positiveteststrategy
possibilityeffect:gamblesand;threatsand
post-traumaticstress
poverty
precautionaryprinciple
predictability,insensitivityto
predictions and forecasts; baseline; clinical vs. statistical; disciplining; of experts, see
expertintuition;extreme,value of; formulasfor,seeformulas;increasingaccuracy
in; low-validity environments and; nonregressive; objections to moderating;
optimistic bias in; outside view in; overconfidence in; planning fallacy and; short-
termtrendsand;valid,illusionof;seealsoprobability
preferencereversals;unjust
premonition,useofword
premortem
pretentiousnesslanguage
pricingpolicies
priming;anchoringas
t=”-5%”>
PrincetonUniversity
probability; base rates in, see base rates; decision weights and, see decision weights;
definitions of; and disciplining intuition; less-is-more pattern and; Linda problem
and;overestimationof;plausibilityand;andpredictingbyrepresentativeness;prior,
insensitivity to; professional stereotypes and; of rare events, see rare events;
representativenessand,seerepresentativeness;similarityand;subjective;assum-like
variable;seealsopredictionsandforecasts
probabilityneglect
ProceedingsoftheNationalAcademyofSciences
professionalstereotypes
professorialcandidates
prospecttheory;inAlbertandBenproblem;blindspotsof;cumulative;decisionweights
andprobabilitiesin;fourfoldpatternin;framesand;graphoflossesandgainsin;loss
aversionin;referencepointsin
“ProspectTheory:AnAnalysisofDecisionUnderRisk”(KahnemanandTversky)
prototypes
psychiatricpatients
psychologicalimmunesystem
psychology,teaching
psychopathiccharm
psychophysics
psychotherapists
pundits;seealsoexpertintuitionpunishments:altruistic;rewardsand;self-administered
pupildilation
questionnaireandgiftexperiments
questions;substitutionof,seesubstitution
Rabin,Matthew
radiologists
rafters,skilled
railprojects
randomnessandchance;misconceptionsofRandomWalkDownWallStreet,A(Malkiel)
rareevents;overestimationof;regretand
rational-agentmodel
rationality
RationalityandtheReflectiveMind(Stanovich)
“>rats
Reagan,Ronald
reciprocalpriming
recognition
recognition-primeddecision(RPD)model
Redelmeier,Don
referenceclassforecasting
regressiontothemean;causalinterpretationsand;correlationand;difficultyingrasping;
two-systemsviewof
“RegressiontowardsMediocrityinHereditaryStature”(Galton)
regret
religion
rememberingself
RemoteAssociationTest(RAT)
reorganizationsincompanies
repetition
representativeness;baseratesand;seealsobaserates;inLindaproblem;predictingby;
professionalstereotypesand;sinsof;inTomWproblem
research:artifactsin;hypothesistestingin;optimismin
resemblance;inpredictions
resilience
responsibility
retrievabilityofinstances
reversals;unjust
rewards;self-administered
Rice,Condoleezza
risk assessment; aggregation and; broad framing in; decision weights in, see decision
weights;denominatorneglectand;byexperts;andformatofriskexpression;fourfold
patternin;forhealthrisks;hindsightbiasand;lawsandregulationsgoverning;loss
aversion in; narrow framing in; optimistic bias and; policies for; possibility effect
and;precautionaryprincipleand;probabilityneglectand;publicpoliciesand;small
risksand;oftechnologies;terrorismand;seealsogambles
riskaversion
riskseeking
“RobustBeautyofImproperLinearModelsinDecisionMaking,The”(Dawes)
Rosett,Richard
Rosenzweig,Philip
RoyalDutchShell
RoyalInstitution
Rozin,Paul
<Philip
Rumsfeld,Donald
RussellSageFoundation
Russia
SaddamHussein
sadness
safety;healthrisksand;healthviolationpenaltiesand;precautionaryprincipleand
samples, sampling: accidents of; and bias of confidence over doubt; law of large
numbers;lawofsmallnumbers;sizeof;small,exaggeratedfaithin
Samuelson,Paul
SanFranciscoExploratorium
Savage,Jimmie
SaveMoreTomorrow
Schelling,Thomas
Schkade,David
schoolsize
Schwarz,Norbert
Schweitzer,Maurice
Science
ScientificAmerican
scientificcontroversies
scientificresearch:artifactsin;hypothesistestingin;optimismin
ScottishParliament
self-control
self-criticism
Seligman,Martin
selves;experiencing;remembering
sets
Shafir,Eldar
similarityjudgments
Simmel,Mary-Ann
Simon,Herbert
Simons,Daniel
Simpson,O.J.
singleevaluations;jointevaluationsvs.
skijumpevent
skills; acquisition of; environment of; feedback and practice in; illusions of; in stock-
picking
Slovic,Paul
Slovic,Roz
slowthinking
<=“0>
smiles,infacereading
smiling;availabilityheuristicand
Smith,Vernon
socializing
socialscience
Soll,Jack
somaticmarkerhypothesis
soul
SourcesofPower(Klein)
SovietUnion
Spinoza,Baruch
SportsIllustrated
Stalin,Joseph
Standard&Poor’s(S&P)
StanfordUniversity
Stanovich,Keith
statisticsandstatisticalthinking;andaccidentsofsampling;baseratesand,seebaserates;
Bayesian;andbiasofconfidenceoverdoubt;causesand;chancein;decidingonsize
ofsample;extremeoutcomesand;faithinsmallsamples;lawoflargenumbers;law
ofsmallnumbers;samplesizedecisionsand;seealsoprobability
statusquo,defending
Steiger,JamesH.
stereotypes;causal;aboutprofessions
Stevethelibrarian
stockmarket
stockpicking
stockportfolios
stocktrading,insider
Stone,Arthur
stories,life
St.Petersburgparadox
Strack,Fritz
strangers,assessmentof
StrangerstoOurselves(Wilson)
Streep,Meryl
strength,assessmentsof
structuredsettlements
StumblingtoHappiness(Gilbert)
substitution;andmoodheuristicforhappiness;and3-Dheuristic
success,uot
sum-likevariables
sunk-costfallacy
Sunstein,Cass
SuperBowl
supplyanddemand
surgeons
Surowiecki,James
surprise
surveyandgiftexperiments
survival-mortalityexperiment
symbols
System1;characteristicsof;conflictbetweenSystem2and
System2;conflictbetweenSystem1and;lazinessof
Taleb,Nassim
talent
tasksets
taskswitching
Tate,Geoffrey
taxes;childexemptionsand
temperament
temptation
Tenet,George
terrorism
Tetlock,Philip
Thaler,Richard
theory-inducedblindness
therapists
thinkinglikeatrader
Thomas,Lewis
threats;possibilityeffectand
3-Dheuristic
tickets;buyingandsellingof;sunkcostin
time;useof
timepressure
Todorov,Alex
tokenexperiment
TomWproblem
“TradingIsHazardoustoYourWealth”(BarberandOdean)
transactionsandtrades
Traviata,La(Verdi)
Truman,Harry
trustworthiness,assessmentsof
truth,illusionsof
Tversky,Amos
understanding,illusionof
uniquecases
UniversityCollegeLondon
UniversityofCaliforniaatBerkeley
UniversityofChicago
UniversityofMichigan
UniversityofMinnesota
UniversityofOregon
unlikelyevents,seerareeventsunknownunknowns
utility;decision;experienced;indifferencemapand;injectionpuzzleand;meaningsof
utilitytheory;certaintyeffectand;decisionweightsandprobabilitiesin
vacations
vaccines
validity:ofclinicalvs.statisticalpredictions;evaluating;illusionof
Vallone,Robert
value;seealsoutilityVancouverIsland
Venndiagrams
venturecapitalists
victimcompensation
vividness;ofoutcomes;ofprobabilities
vocabulary:ofgirlsvs.boys;simplevs.pretentious
Vohs,Kathleen
vomit,effectofword
VonNeumann,John
voting
Wainer,Howard
walking
wars
WashingtonPost,The
wealth,seemoneyandwealth
weather
Weber,Ernste>
weightandpianoplaying,measuring
Weiner,Howard
well-being; climate and; defining; disposition for; duration weighting and; see also
happiness
West,Richard
what you see is all there is (WYSIATI); confidence and; curriculum team and; Julie
problem and; optimistic bias and; premortem and; professorial candidate problem
and;soldiers’performanceand;TomWproblemand
wheeloffortune
“wicked”environments
Wilson,Timothy
Wimbledontournament
wine
WinterOlympics
WisdomofCrowds,The(Surowiecki)
witnesses’evidence
Woods,Tiger
words:complexvs.simple;emotionally-loaded
WorldCup
WorldWarII
worry
WYSIATI,seewhatyouseeisallthereis
X-rays
Xu,Jing
Yaleexamproblem
YomKippurWar
Zajonc,Robert
Zamir,Eyal
Zeller,Kathryn
Zweig,Jason
Zwerling,Harris
P
Farrar,StrausandGiroux
18West18thStreet,NewYork10011
Copyright©2011byDanielKahneman
Allrightsreserved
Grateful acknowledgment is made for permission to reprint the following previously
publishedmaterial:“JudgmentUnderUncertainty:HeuristicsandBiases”fromScience,
NewSeries,Vol.185,No.4157,copyright©1974byAmosTverskyandDan”0%”te>X-
raysScience.“Choices,Values,andFrames”fromTheAmericanPsychologist,copyright
©1983byDanielKahnemanandAmosTversky.ReprintedbypermissionoftheAmerican
PsychologicalAssociation.
Gratefulacknowledgmentismadeforpermissiontoreprintthefollowingimages:Image
courtesy of Paul Ekman Group, LLC. Image from “Cues of Being Watched Enhance
Cooperation in a Real-World Setting” by Melissa Bateson, Daniel Nettle, and Gilbert
Roberts,BiologyLetters(2006);reprintedbypermissionofBiologyLetters.Imagefrom
Mind Sights by Roger N. Shepard (New York: W.H. Freeman and Company, 1990);
reprinted by permission of Henry Holt and Company. Image from “Human Amygdala
Responsivity to Masked Fearful Eye Whites” by Paul J. Whalen et al., Science 306
(2004).ReprintedbypermissionofScience.
LibraryofCongressCataloging-in-PublicationData
Kahneman,Daniel,1934–
Thinking,fastandslow/DanielKahneman.—1sted.
p.cm.
Includesbibliographicalreferencesandindex.
ISBN:978-0-3742-7563-1
1.Thoughtandthinking.2.Decisionmaking.3.Intuition.4.Reasoning.I.Title.
BF441.K2382011
153.4‘2—dc23
2011027143
www.fsgbooks.com
P
*ThisarticleoriginallyappearedinScience,vol.185,1974.Theresearchwassupported
by the Advanced Research Projects Agency of the Department of Defense and was
monitored by the Office of Naval Research under contract N00014-73-C-0438 to the
OregonResearchInstitute,Eugene.Additionalsupportforthisresearchwassr”0%”wid
provided by the Research and Development Authority of the Hebrew University,
Jerusalem,Israel.
P
*ThisarticlewasoriginallypresentedasaDistinguishedScientificContributionsAward
addressattheAmericanPsychologicalAssociationmeeting,August1983.Thisworkwas
supported by grant NR 197-058 from the U.S. Office of Naval Research. Originally
publishedinAmericanPsychologist,vol.34,1984.
P
P
